Sep. 01, 2025
Agricultural
The pear is a nutrient-rich fruit with high economic and nutritional value, widely cultivated around the world (Seo et al., ). China has been the largest producer and consumer of fruits globally, with orchard area and production continuously increasing (Zhang et al., ). Fruit harvesting has become one of the most time-consuming and labor-intensive processes in fruit production (Vrochidou et al., ). In the complex orchard environment, accurate fruit detection is essential for achieving orchard automation and intelligent management (Bharad and Khanpara, ; Chen et al., ). Currently, pear harvesting mainly relies on manual labor, which is inefficient. Additionally, with the aging population and labor shortages, the cost of manual harvesting is rising, making the automation of pear fruit harvesting an urgent problem to address. In recent years, researchers have been focusing on mechanized and intelligent fruit harvesting technologies (Parsa et al., ). However, in the unstructured environment of orchards, fruits are often occluded by branches and leaves, and their growth orientations vary, which affects the accuracy of detection and localization, posing significant challenges to automated fruit harvesting (Tang et al., ).
For more information, please visit our website.
Traditional image processing methods for detecting fruit targets require manually designed features, such as color features, shape features, and texture features (Liu and Liu, ). These methods then combine machine learning algorithms with the manually designed features to detect fruits, but detection accuracy can be easily affected by subjective human factors, and detection efficiency is low (Dhanya et al., ).
In recent years, with the development of image processors (GPUs) and deep learning technologies, significant progress has been made in the field of object detection. Algorithms such as Faster R-CNN (Ren et al., ) and SSD (Liu et al., ) have demonstrated excellent performance in general tasks. However, these methods still face challenges in real-time processing or small object detection. The YOLO series such as YOLOv5 (Horvat et al., ), YOLOv6 (Li et al., ), YOLOv7 (Wang et al., ), YOLOv8 (Sohan et al., ), YOLOv9 (Wang et al., ), YOLOv10 (Alif and Hussain, ), YOLOv11 (Khanam and Hussain, ) has shown improvements in both speed and accuracy, leading many researchers to utilize YOLO algorithms for fruit detection research. Liu et al. () proposed a new lightweight apple detection algorithm called Faster-YOLO-AP based on YOLOv8. The results showed that Faster-YOLO-AP reduced its parameters and FLOPs to 0.66 M and 2.29G, respectively, with an mAP@0.5:0.95 of 84.12%. Zhu et al. () introduced an improved lightweight YOLO model (YOLO-LM) based on YOLOv7-tiny for detecting the maturity of tea oil fruits. The precision, recall, mAP@0.5, parameters, FLOPs, and model size were 93.96%, 93.32%, 93.18%, 10.17 million, 19.46 G, and 19.82 MB, respectively. Wei et al. () proposed a lightweight tomato maturity detection model named GFS-YOLOv11, which improved precision, recall, mAP@0.5, and mAP@0.5:0.95 by 5.8%, 4.9%, 6.2%, and 5.5%, respectively. Tang et al., addressed the issue of low detection accuracy and limited generalization capabilities for large non-green mature citrus fruits under different ripeness levels and varieties, proposing a lightweight real-time detection model for unstructured environments—YOLOC-tiny. Sun et al. () focused on efficient pear fruit detection in complex orchard environments and proposed an effective YOLOv5-based model—YOLO-P—for fast and accurate pear fruit detection. However, in complex, unstructured orchard environments, factors such as varying lighting conditions, occlusions, and fruit overlaps still affect recognition accuracy and generalization capabilities. Additionally, existing models often suffer from high computational complexity and excessive parameters, making them difficult to deploy on resource-constrained mobile or embedded devices. To address these challenges, researchers have been committed to designing high-precision, fast detection models that meet the requirements for real-time harvesting.
Current research on pear fruit detection has made some progress. Ren et al. () proposed the YOLO-GEW network based on YOLOv8 for detecting “Yulu Xiang” pear fruits in unstructured environments, achieving a 5.38% improvement in AP. Zhao et al. () developed a high-order deformation-aware multi-object search network (HDMNet) based on YOLOv8 for pear fruit detection, with a detection accuracy of 93.6% in mAP@0.5 and 70.2% in mAP@0.75. Lu et al. () introduced the ODL Net algorithm for detecting small green pear fruits, achieving detection accuracies of 56.2% and 65.1% before and after fruit thinning, respectively. Shi et al. () proposed an improved model, YOLOv9s-Pear, based on the lightweight YOLOv9s model, enhancing the accuracy and efficiency of red-skinned young pear recognition. The model achieved precision, recall, and AP rates of 97.1%, 97%, and 99.1%, respectively. The aforementioned studies primarily focus on single pear fruit detection during maturity or young fruit stages. However, in practical harvesting scenarios, considerations such as robotic arm picking strategies and path planning are also crucial (Wang et al., ). The picking strategy and path planning of robotic arms are closely related to the fruit’s growth position. Detailed classification of fruit location information enables harvesting robots to adapt flexibly to varying environmental conditions, dynamically adjusting path planning and grasping strategies to ensure efficient and precise harvesting operations. This enhances the system’s flexibility and robustness in complex scenarios (Nan et al., ).
Based on the aforementioned background, this paper proposes a lightweight intelligent pear orchard fruit detection method, AHG-YOLO, using YOLOv11n as the base model. First, the traditional sampling method in the YOLOv11n backbone and neck networks is replaced with ADown to reduce computational complexity while improving detection accuracy. Next, a new detection head structure is developed using the “shared” concept and group convolution to further lighten the model without compromising detection performance. Finally, the CIoU loss function in YOLOv11n is replaced with GIoU to enhance the model’s accuracy and fitting capability. The improved model not only maintains high recognition accuracy but also reduces the model size and computational cost, making it easier to deploy on mobile devices. This provides technical support for optimizing robotic picking paths and meets the demands of intelligent harvesting in pear orchards.
The Hongxiangsu pear, known as the “king of all fruits,” is a hybrid descendant of the Korla fragrant pear and the goose pear, and is a late-maturing, storage-resistant red-skinned pear variety. The fruit is spindle-shaped, weighing an average of 220 grams, with a maximum weight of 500 grams. The fruit surface is smooth and clean, with a bright red color. The flesh is white, fine-grained, sweet, and aromatic, with medicinal properties such as clearing heat, moisturizing the lungs, relieving cough, quenching thirst, and aiding in alcohol detoxification. It also has health benefits for conditions such as hypertension, high cholesterol, and arteriosclerosis. This study focuses on the Hongxiangsu pear, and data was collected from the Modern Agricultural Industry Technology System Demonstration Base of the Fruit Tree Research Institute at Shanxi Agricultural University, located in Taigu District, Jinzhong City, Shanxi Province (112°32’E, 37°23’N). Considering that the harvesting robotic arm needs to adapt to the complex environment of the orchard during the harvesting process, pear images were captured from various angles, distances, and time periods using a Vivo Z3i smartphone. A total of 1,000 pear images were collected, and unusable images were filtered out, leaving 734 usable images. The complex orchard environment includes scenarios such as single fruit, multiple fruits, cloudy weather, overlapping fruits, and branches and leaves obstructing the view. Some sample images are shown in Figure 1.
Figure 1To improve the robustness and generalization ability of the pear object detection model, image sample data needs to be augmented. In this study, various augmentation techniques, including adding salt-and-pepper noise, image sharpening, affine transformation, and brightness adjustment, are randomly combined to enhance the images. After data augmentation, the total number of pear samples is . The dataset is split into training set ( images), validation set (293 images), and test set (588 images) with a ratio of 7:1:2. Some of the augmented data samples are shown in Figure 2.
Figure 2In natural environments, pear fruits are often obstructed by leaves or branches, and fruits can occlude each other, posing significant challenges for robotic harvesting. To improve harvesting efficiency, the harvesting robot can adopt different strategies when encountering pears in various scenarios during the harvesting process. For example, for an unobstructed target, path planning is relatively simple, and conventional path planning and grabbing tasks can be directly applied. When the target is partially occluded, path planning needs to consider how to navigate around the obstruction or adjust the grabbing angle. In environments with dense fruits, where occlusion and overlap of multiple fruits are concerns, multi-object path planning algorithms can be used to devise the optimal path (Gao, ; Yang et al., ). Therefore, based on the growth loci characteristics, the fruits are systematically categorized into three distinct classes in this study. The schematic of the three categories of pears is shown in Figure 3. The first class represents fruits that are not obstructed (referred to as NO). The second class represents fruits that are occluded by branches or leaves (referred to as OBL). The third class represents fruits that are in close contact with other fruits but are not occluded by branches or leaves (referred to as FCC). This classification standard is based on the classification criteria proposed by Nan et al. () for pitaya fruits.
Figure 3The pear fruits in the images were annotated using rectangular bounding boxes in Labeling (Tzutalin, ) software, categorized into three classes (NO, OBL, and FCC) according to the predefined classification criteria. The annotations were formatted in YOLO style and ultimately saved as.txt files. Upon completion of the annotation process, the distribution of different categories across the final training set, validation set, and test set is shown in Table 1.
Table 1The YOLOv11 network introduces two innovative modules, C3k2 and C2PSA, as shown in Figure 4, which further enhance the network’s accuracy and speed. However, in unstructured environments such as orchards, when fruits are severely occluded, overlapping, or when the fruit targets are small, the YOLOv11 network is prone to missing or misdetecting targets. To enhance the accuracy and robustness of pear detection algorithms in unstructured environments, this paper improves the YOLOv11n model. The architecture of the improved model is shown in Figure 5. First, in both the backbone and head networks, the downsampling method is replaced with ADown (Wang et al., ), enabling the model to capture image features at higher levels, enhancing the feature extraction capability of the network and reducing computational complexity. Then, a lightweight detection head, Detect_Efficient, is designed, which further reduces the computational load by sharing the detection head and incorporating group convolution, while improving the network’s feature extraction capacity. Finally, the CIou loss function of YOLOv11 is replaced with GIoU (Jiang et al., ), which reduces the impact of low-quality samples and accelerating the convergence of the network model. The proposed improvements are named AHG-YOLO, derived from the first letters of the three improvement methods: ADown, Head, and GIoU. The AHG-YOLO model effectively improves pear detection performance and better adapts to the detection needs of small targets, occlusion, and fruit overlap in the complex natural environment of pear orchards.
Figure 4Figure 5The ADown module in YOLOv9 is a convolutional block for downsampling in object detection tasks. As an innovative feature in YOLOv9, it provides an effective downsampling solution for real-time object detection models, combining lightweight design and flexibility. In deep learning models, downsampling is a common technique used to reduce the spatial dimensions of feature maps, enabling the model to capture image features at higher levels while reducing computational load. The ADown module is specifically designed to perform this operation efficiently with minimal impact on performance.
The main features of the ADown module are as follows: (1) Lightweight design: The ADown module reduces the number of parameters, which lowers the model’s complexity and enhances operational efficiency, especially in resource-constrained environments. (2) Information preservation: Although ADown reduces the spatial resolution of feature maps, its design ensures that as much image information as possible is retained, allowing the model to perform more accurate object detection. (3) Learnable capabilities: The ADown module is designed to be learnable, meaning it can be adjusted according to different data scenarios to optimize performance. (4) Improved accuracy: Some studies suggest that using the ADown module not only reduces the model size but also improves object detection accuracy. (5) Flexibility: The ADown module can be integrated into both the backbone and head of YOLOv9, offering various configuration options to suit different enhancement strategies. (6) Combination with other techniques: The ADown module can be combined with other enhancement techniques, such as the HWD (Wavelet Downsampling) module, to further boost performance. The ADown network structure is shown in Figure 6.
Figure 6By introducing the ADown module into YOLOv9, a significant reduction in the number of parameters can be achieved, while maintaining or even improving object detection accuracy. Consequently, this study explores the integration of the ADown module into the YOLOv11 network structure to further enhance detection performance.
The detection decoupled head structure of YOLOv11n is shown in Figure 7. The extracted feature map is passed through two branches. One branch undergoes two 3×3 convolutions followed by a 1×1 convolution, while the other branch undergoes two depth wise separable convolutions (DWConv), two 3×3 convolutions, and a 1×1 convolution. These branches are used to independently predict the bounding box loss and the classification function.
Figure 7In YOLOv11, there are three of the aforementioned decoupled head structures, which perform detection on large, medium, and small feature maps. However, 3×3 convolutions, while increasing the channel depth, lead to a significant increase in the number of parameters and floating-point operations (Shafiq and Gu, ). Therefore, this study aims to implement a lightweight design for YOLOv11’s detection head while maintaining detection accuracy:
(1) Introducing Group Convolutions to Replace 3×3 Convolutions.
Group Convolution is a convolution technique used in deep learning primarily to reduce computation and parameter quantities while enhancing the model’s representational power. Group convolution works by dividing the input feature map and convolution kernels into several groups. Each group performs its convolution operation independently, and the results are then merged. This process reduces the computation and parameter quantities while maintaining the same output size.
In traditional convolution operations, the convolution is applied across every channel of the input feature map. Assuming the input feature map has dimensions Cin×H×W (Cin is the number of input channels, H is the feature map height, and W is the feature map width), and the convolution kernel has dimensions Cout×Cin×k×k (Cout is the number of output channels and k×k is the spatial dimension of the kernel), the computation for a single convolution operation is: Cout×Cin×k×k×H×W.
In group convolution, the input channels are divided into g groups, and independent convolution operations are performed within each group. In this case, for each group, the number of input channels becomes Cin/g, and the computation becomes: Cout×Cin/g×k×k×H×W.
Group convolution can greatly reduce the number of parameters, enhance the model’s representational power, and avoid overfitting. Therefore, the 3×3 convolutions in the detection head are replaced with group convolutions.
(2) Shared Convolution Parameters.
To further reduce the parameters and computation of the detection head, the two branch inputs of the detection head share two group convolutions, named Detect_Efficient, with the structure shown in Figure 8. By sharing the same convolution kernel weights during loss calculation, redundant computation of similar feature maps is avoided, which further reduces the computation and effectively improves computational efficiency, accelerating the entire model inference process.
Figure 8The boundary box loss function is an important component of the object detection loss function. A well-defined boundary box loss function can significantly improve the performance of object detection models. In YOLOv11, CIoU is used as the regression box loss function. Although CIoU improves upon GIoU by introducing center distance and aspect ratio constraints, the additional constraints introduced by CIoU might lead to overfitting or convergence difficulties in orchard data collection, where there is a large variation in target size (due to close and distant objects) and where the aspect ratio differences of pear fruit bounding boxes are not significant. Moreover, compared to GIoU, the calculation of the aspect ratio parameter v in CIoU is relatively more complex (Zheng et al., ), resulting in higher computational costs during training and slower model convergence. Therefore, this study replaces CIoU with the GIoU loss function. The GIoU loss function is used in object detection to measure the difference between the predicted and ground truth boxes, addressing the issue where traditional IoU fails to provide effective gradient feedback when the predicted box and the ground truth box do not overlap. This improves the model’s convergence and accuracy. GIoU loss not only considers the overlapping region between boxes but also takes into account the spatial relationship between the boxes by introducing the concept of the minimal enclosing box. This allows the model to learn the shape and position of the boxes more accurately, ultimately enhancing the performance of object detection.
The experimental environment for this study runs on the Windows 10 operating system, equipped with 32 GB of memory and an NVIDIA GeForce RTX GPU, with an Intel(R) Core(TM) i7-F @2.10GHz processor. The deep learning framework used is PyTorch 2.0.1, with CUDA 11.8 and CUDNN 8.8.0.
The network training parameters are set as follows: The image input size is 640 × 640, and the batch size is set to 32; the maximum number of iterations is 200. The optimizer is SGD, with the learning rate dynamically adjusted using a cosine annealing strategy. The initial learning rate is set to 0.01, the momentum factor is 0.937, and the weight decay coefficient is 0..
Object detection models should be evaluated using multiple metrics to provide a comprehensive assessment of their performance. To evaluate the performance of ADG-YOLO, seven metrics are used: precision, recall, average precision (AP), mean average precision (mAP), number of parameters, model size, and GFLOPs. These metrics offer a well-rounded evaluation of ADG-YOLO’s performance in the multi-category pear fruit detection task within the complex environment of a pear orchard. They reflect the model’s performance across various dimensions, including accuracy, recall, speed, and efficiency. The formulas for calculating the relevant performance metrics are provided, as shown in Equations 1-4.
Where TP represents the number of true positive samples that the model correctly predicted as positive, FP represents the number of false positive samples that the model incorrectly predicted as positive, and FN represents the number of false negative samples that the model incorrectly predicted as negative. AP refers to the area under the Precision-Recall (P-R) curve, while mAP refers to the mean value of the AP for each class.
To evaluate the effectiveness and feasibility of the proposed AHG-YOLO model in detecting pear fruits with no occlusion, partial occlusion, and fruit overlap, an ablation experiment was conducted based on the YOLOv11n model. Each improvement method and the combination of two improvement methods were added to the YOLOv11n model and compared with the AHG-YOLO model. In the experiment, the hardware environment and parameter settings used for training all models remained consistent. Table 2 shows the ablation experiment results of the improved YOLOv11n model and the AHG-YOLO model on the test set. After introducing the ADown downsampling module to enhance the feature extraction capability of the YOLOv11 network, the model’s precision, recall, AP, and mAP@0.5:0.95 increased by 2.2%, 2.6%, 1.8%, and 2.1%, respectively. The model’s parameter count decreased by 18.6%, GFLOPs decreased by 15.9%, and model size decreased by 17.3%. This indicates that the ADown module can effectively improve the pear object detection accuracy. After introducing the EfficientHead detection head, although the model’s precision, recall, and AP decreased slightly, mAP@0.5:0.95 increased by 0.1%, the model’s parameter count reduced by 10.4%, GFLOPs reduced by 19.0%, and model size decreased by 9.62%. This suggests that EfficientHead plays a significant role in model lightweighting. As shown in Table 2, after introducing the ADown module and GIoU, although the model’s parameter count increased, precision, recall, mAP@0.5, and mAP@0.5:0.95 increased by 1.5%, 1.7%, 0.4%, and 1.6%, respectively. After introducing the ADown module and EfficientHead, precision, recall, mAP@0.5, and mAP@0.5:0.95 increased by 2.3%, 2.2%, 1.7%, and 2.5%, and the model’s parameter count, GFLOPs, and model size all decreased. Additionally, after introducing EfficientHead and GIoU, recall, mAP@0.5, and mAP@0.5:0.95 all increased compared to their individual introduction, without increasing the parameter count. Finally, the proposed AHG-YOLO network model outperforms the original YOLOv11 model, with precision, recall, mAP@0.5, and mAP@0.5:0.95 improving by 2.5%, 3.6%, 2.3%, and 2.6%, respectively. Meanwhile, GFLOPs are reduced to just 4.7, marking a 25.4% decrease compared to the original YOLOv11n, the parameter count decreased by 16.9%, and the model size is only 5.1MB.
Table 2According to the data in Table 2, the mAP@0.5 of YOLOv11-A reached 93.6%, an improvement over the baseline model YOLOv11n. However, when H or G were added individually, the mAP@0.5 dropped to 90.7% and 90.8%, respectively. When combined with the A module, the mAP values increased again. The reasons for this can be analyzed as follows: The ADown module significantly improves baseline performance by preserving discriminative multi-scale features through adaptive downsampling. The EfficientHead method reduces model parameters and computational load compared to the baseline model, but the simplified model structure leads to information loss and a decrease in detection accuracy. GIoU performs poorly on bounding box localization in raw feature maps, resulting in a drop in detection accuracy. When combined with ADown, the ADown module optimizes the features, providing better input for the subsequent EfficientHead and GIoU, thus leveraging the complementary advantages between the modules. The optimized features from ADown reduce the spatial degradation caused by EfficientHead, maintaining a mAP@0.5 of 93.5%, while reducing GFLOPs by 11.3%. ADown’s noise suppression allows GIoU to focus on key geometric deviations, improving localization robustness. The synergy of all three modules achieves the best accuracy-efficiency balance (94.1% mAP@0.5, 4.7 GFLOPs), where ADown filters low-level redundancies, EfficientHead enhances discriminative feature aggregation, and GIoU refines boundary precision. This analysis shows that H and G are not standalone solutions, they require the preprocessing from ADown to maximize their effectiveness.
Figures 9 and 10 show the performance of AHG-YOLO compared to YOLOv11n during the training process. From Figures 9, 10, it can be seen that during 200 training iterations, the proposed AHG-YOLO achieves higher detection accuracy and obtains lower loss values compared to YOLOv11n. This indicates that the AHG-YOLO network model can effectively improve the detection accuracy of pears in unstructured environments and reduce the false detection rate.
Figure 9Figure 10The Grad-CAM (Selvaraju et al., ) method is used to generate heatmaps to compare the feature extraction capabilities of the YOLOv11n model and the AHG-YOLO model in complex scenarios such as overlapping fruits, small target fruits, and fruit occlusion, as shown in Figure 11. Figure 11 shows that the AHG-YOLO model exhibits better performance in complex scenarios. The specific quantitative results comparison can be found in Section 3.3.
Figure 11To further validate the detection performance, experiments were conducted on the test dataset for both the YOLOv11n model and the AHG-YOLO model. The detection results are shown in Figure 12, where the red circles represent duplicate detections and the yellow circles represent missed detections. By comparing Figures 12A, B, it can be observed that YOLOv11n has one missed detections. By comparing Figures 12C, D, it can be seen that YOLOv11n has one duplicate detections. By comparing Figures 12E, F, it can be seen that YOLOv11n has two duplicate detections and one missed detections. This demonstrates that AHG-YOLO can accurately perform multi-class small object detection and classification in complex environments, exhibiting high accuracy and robustness, and effectively solving the pear detection problem in various scenarios within complex environments.
Figure 12Figure 13 shows the AP results for multi-category detection for occluded pear fruits in complex orchard scenes by different networks on the test set. Table 3 presents the specific detection results of AHG-YOLO and YOLOv11n for different categories of pear targets on the test set. From Figure 13 and Table 3, it can be observed that the base YOLOv11 network performs best in detecting NO fruit, with an AP value of 93.4%, but performs relatively poorly when detecting FCC and OBL fruits. The proposed AHG-YOLO model improves the AP for detecting FCC fruits by 2.6%, reaching 93.4%, improves the AP for detecting OBL fruits by 2.4%, reaching 93.5%, and improves the AP for detecting NO fruits by 1.9%, reaching 95.3%. This indicates that the proposed method is highly effective for fruit target detection in complex environments, demonstrating both excellent accuracy and robustness.
Figure 13Table 3AHG-YOLO was compared with other mainstream object detection networks, and the detection results on the test set are shown in Table 4. The experimental results of all models indicate that YOLOv9c achieves the highest precision, mAP@0.5, and mAP@0.5:0.95 among all models. However, the YOLOv9c model has excessively large parameters, GFLOPs, and model size, making it unsuitable for real-time detection in harvesting robots. AHG-YOLO’s mAP@0.5 surpasses that of Faster R-CNN, RTDETR, YOLOv3, YOLOv5n, YOLOv7, YOLOv8n, YOLOv10n, and YOLOv11n by 15.1%, 0.9%, 2.4%, 3.9%, 12.6%, 3.4%, 5.2%, and 2.3%, respectively. In terms of precision, recall, mAP@0.5:0.95, and GFLOPs, AHG-YOLO also shows advantages. Therefore, based on a comprehensive comparison of all metrics, AHG-YOLO is better suited for pear target detection tasks in complex environments.
Table 4YOLO series detection algorithms are widely used in fruit detection due to their high detection accuracy and fast detection speed. These algorithms have been applied to various fruits, such as tomatoes (Wu H, et al., ), kiwifruits (Yang et al., ), apples (Wu M, et al., ), achieving notable results. Researchers have always been focused on designing lightweight algorithms, and this is also true for pear fruit target detection. Tang et al. () proposed a pear target detection method based on an improved YOLOv8n for fragrant pears. Using their self-built fragrant pear dataset, they improved the F0.5-score and mAP by 0.4 and 0.5 percentage points compared to the original model, reaching 94.7% and 88.3%, respectively. Li et al. () introduced the advanced multi-scale collaborative perception network YOLOv5sFP for pears detection, achieving an AP of 96.12% and a model size of 50.01 MB.
While these studies have achieved remarkable results, they did not address the practical needs of robotic harvesting, as they focused solely on detecting a single class of pear fruits. This study takes into account the detection requirements for robotic harvesting, categorizing pear fruits in orchards into three groups (ON, OBL, FCC) to enable the harvesting robot to develop different harvesting strategies based on conditions of no occlusion, branch and leaf occlusion, and fruit overlap, thus improving harvesting efficiency. Compared to commonly used detection models, the AHG-YOLO proposed in this study achieves the highest detection accuracy in complex orchard environments, with an mAP of 94.1%.
Figure 14 shows three examples of detection errors when using AHG-YOLO for multi-category detection of occluded pear fruits. The potential causes of these errors are as follows: (1) In cloudy, dim lighting conditions, when fruits are tightly clustered and located at a distance, the fruit targets appear small, making feature extraction challenging. This leads to repeated detection of FCC fruits, as seen in the lower right red circle of Figure 14A. Additionally, the dim lighting causes the occluded pear’s features to resemble those of the leaves, resulting in the model mistakenly detecting leaves as OBL fruits, as shown in the upper left red circle of Figure 14A. (2) When the target is severely occluded, the model struggles with feature extraction, which may lead to either missed detections or repeated detection, as shown in Figure 14B. The yellow bounding box indicates a missed detection, and the red circle indicates a repeated detection. (3) Detecting FCC fruits is particularly challenging because the fruits are often clustered together, making it difficult to distinguish between them. Furthermore, the fruit bags sometimes interfere with the detection process, causing errors, as seen in Figure 14C, where the bag is incorrectly detected as an FCC fruit.
Figure 14To enhance the accuracy of AHG-YOLO in detecting multi-category detection for occluded pear fruits, the following measures can be taken: (1) Increase the number of samples that are prone to detection errors, such as FCC and OBL class samples, to diversify the dataset and improve the model’s detection capability in complex environments. (2) Further refine the model’s feature extraction capability, particularly for detecting small targets.
Although the AHG-YOLO model has some limitations in detecting multi-category detection for occluded pear fruits, it achieves an overall detection mAP of 94.1%, which meets the fruit detection accuracy requirements for orchard automation in harvesting. This provides crucial technical support for robotic pear harvesting in orchards. The AHG-YOLO model will be applied to the visual detection system of pear fruit-picking robots to validate its reliability.
This paper proposes the AHG-YOLO network model for multi-category detection of occluded pear fruits in complex orchard scenes. Using YOLOv11n as the base model, the ADown downsampling method, lightweight detection head, and GIoU loss function are integrated to enhance the network’s feature extraction capability and reduce the model’s complexity, making it suitable for real-time harvesting applications. The conclusions are as follows:
(1) Experimental results in complex pear orchard environments demonstrate that the mAP of AHG-YOLO for multi-category detection for occluded pear fruits is 94.1%, with the AP for FCC, OBL, and NO fruits being 93.4%, 93.5%, and 95.3%, respectively. Compared to the base YOLOv11n network, precision, recall, mAP@0.5, and mAP@0.5:0.95 improved by 2.5%, 3.6%, 2.3%, and 2.6%, respectively. Additionally, GFLOPs are reduced to 4.7, representing a 25.4% decrease compared to the original YOLOv11n, while the number of parameters is reduced by 16.9%, and the model size is just 5.1MB.
(2) Compared with eight other commonly used object detection methods, AHG-YOLO achieves the highest detection accuracy while maintaining a lightweight design. The mAP@0.5 is 15.1%, 0.9%, 2.4%, 3.9%, 12.6%, 3.4%, 5.2%, and 2.3% higher than Faster R-CNN, RTDETR, YOLOv3, YOLOv5n, YOLOv7, YOLOv8n, YOLOv10n, and YOLOv11n, respectively, thereby meeting the real-time harvesting requirements of orchards.
In summary, the AHG-YOLO model proposed in this paper provides a solid methodological foundation for real-time pear target detection in orchard environments and supports the development of pear-picking robots. Future work will focus on further validating the effectiveness of the method in pear orchard harvesting robots, with ongoing optimization efforts.
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.
NM: Writing – original draft, Writing – review & editing, Conceptualization, Investigation, Software, Supervision, Visualization. YS: Data curation, Investigation, Software, Validation, Visualization, Writing – original draft. CL: Data curation, Visualization, Writing – review & editing. ZL: Data curation, Software, Writing – review & editing. HS: Conceptualization, Funding acquisition, Supervision, Visualization, Writing – original draft, Writing – review & editing.
The author(s) declare that financial support was received for the research and/or publication of this article. This research was supported by the key R&D Program of Shanxi Province (CYJSTX07-23); the Fundamental Research Program of Shanxi Province (No. ).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The author(s) declare that no Generative AI was used in the creation of this manuscript.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Alif, M. and Hussain, M. (). YOLOv1 to YOLOv10: A comprehensive review of YOLO variants and their application in the agricultural domain. arxiv preprint arxiv:.. 13. doi: 10./arXiv..
Crossref Full Text | Google Scholar
Bharad, N. and Khanpara, B. (). Agricultural fruit harvesting robot: An overview of digital agriculture. Plant Arch. 24, 154–160. doi: 10./PLANTARCHIVES..v24.SP-GABELS.023
Crossref Full Text | Google Scholar
Chen, Z., Lei, X., Yuan, Q., Qi, Y., Ma, Z., Qian, S., et al. (). Key technologies for autonomous fruit-and vegetable-picking robots: A review. Agronomy 14, 1-2. doi: 10./agronomy
Crossref Full Text | Google Scholar
Dhanya, V., Subeesh, A., Kushwaha, N., Vishwakarma, D., Kumar, T., Ritika, G., et al. (). Deep learning based computer vision approaches for smart agricultural applications. Artif. Intell. Agric. 6, 211–229. doi: 10./j.aiia..09.007
Crossref Full Text | Google Scholar
Gao, X. (). Research on Path Planning of Apple Picking Robotic Arm Based on Algorithm Fusion and Dynamic Switching [D]. Hebei Agricultural University. doi: 10./d.cnki.ghbnu..
Crossref Full Text | Google Scholar
Horvat, M., Jelečević, L., and Gledec, G. (). “A comparative study of YOLOv5 models performance for image localization and classification,” in Central European Conference on Information and Intelligent Systems. Varazdin, Croatia: Faculty of Organization and Informatics Varazdin. 349–356.
Google Scholar
Jiang, K., Itoh, H., Oda, M., Okumura, T., Mori, Y., Misawa, M., et al. (). Gaussian affinity and GIoU-based loss for perforation detection and localization from colonoscopy videos. Int. J. Comput. Assisted Radiol. Surg. 18, 795–805. doi: 10./s-022--x
PubMed Abstract | Crossref Full Text | Google Scholar
Khanam, R. and Hussain, M. (). Yolov11: An overview of the key architectural enhancements. arxiv preprint arxiv:.. 3–7. doi: 10./arXiv..
Crossref Full Text | Google Scholar
Li, C., Li, L., Jiang, H., Weng, K., Geng, Y., Li, L., et al. (). YOLOv6: A single-stage object detection framework for industrial applications. arxiv preprint arxiv:.. doi: 10./arXiv..
Crossref Full Text | Google Scholar
Liu, Z., Abeyrathna, R., Sampurno, R., Nakaguchi, V., and Ahamed, T. (). Faster-YOLO-AP: A lightweight apple detection algorithm based on improved YOLOv8 with a new efficient PDWConv in orchard. Comput. Electron. Agric. 223, . doi: 10./j.compag..
Crossref Full Text | Google Scholar
Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C., et al. (). “Ssd: Single shot multibox detector,” in Computer Vision–ECCV : 14th European Conference, Amsterdam, The Netherlands, October 11–14, , Proceedings, Part I 14. Cham, Switzerland: Springer International Publishing. 21–37.
Google Scholar
Liu, J. and Liu, Z. (). The vision-based target recognition, localization, and control for harvesting robots: A review. Int. J. Precis. Eng. Manufacturing 25, 409–428. doi: 10./s-023--7
Crossref Full Text | Google Scholar
Lu, Y., Du, S., and Ji, Z. (). ODL Net: Object detection and location network for small pears around the thinning period. Comput. Electron. Agric. 212, . doi: 10./j.compag..
Crossref Full Text | Google Scholar
Nan, Y., Zhang, H., Zeng, Y., Zheng, J., and Ge, Y. (). Intelligent detection of Multi-Class pitaya fruits in target picking row based on WGB-YOLO network. Comput. Electron. Agric. 208, . doi: 10./j.compag..
Crossref Full Text | Google Scholar
Parsa, S., Debnath, B., and Khan, M. (). Modular autonomous strawberry picking robotic system. J. Field Robotics 41, –. doi: 10./rob.
Crossref Full Text | Google Scholar
Ren, S., He, K., Girshick, R., and Sun, J. (). Faster R-CNN: Towards real-time object detection with region proposal networks. IEEE Trans. Pattern Anal. Mach. Intell. 39, –. doi: 10./TPAMI..
PubMed Abstract | Crossref Full Text | Google Scholar
Ren, R., Sun, H., Zhang, S., Wang, N., Lu, X., Jing, J., et al. (). Intelligent Detection of lightweight “Yuluxiang” pear in non-structural environment based on YOLO-GEW. Agronomy 13, . doi: 10./agronomy
Crossref Full Text | Google Scholar
Selvaraju, R., Das, A., Vedantam, R., Cogswell, M., Parikh, D., and Batra, D. (). Grad-CAM: Why did you say that? arxiv preprint arxiv:.. 2-4. doi: 10./arXiv..
Crossref Full Text | Google Scholar
Seo, H., Sawant, S., and Song, J. (). Fruit cracking in pears: its cause and management—a review. Agronomy 12, . doi: 10./agronomy
Crossref Full Text | Google Scholar
Shafiq, M. and Gu, Z. (). Deep residual learning for image recognition: A survey. Appl. Sci. 12, . doi: 10./app
Crossref Full Text | Google Scholar
Shi, Y., Duan, Z., and Qing, S. (). YOLOV9S-Pear: A lightweight YOLOV9S-based improved model for young Red Pear small-target recognition. Agronomy 14, . doi: 10./agronomy
Crossref Full Text | Google Scholar
If you want to learn more, please visit our website Hebei Xingtai.
Featured content:Sohan, M., Sai Ram, T., and Rami Reddy, C. (). A review on yolov8 and its advancements. Int. Conf. Data Intell. Cogn. Inf., 529–545. doi: 10./978-981-99--2_39
Crossref Full Text | Google Scholar
Sun, H., Wang, B., and Xue, J. (). YOLO-P: An efficient method for pear fast detection in complex orchard picking environment. Front. Plant Sci. 13. doi: 10./fpls..
PubMed Abstract | Crossref Full Text | Google Scholar
Tang, Y., Qiu, J., Zhang, Y., Wu, D., Cao, Y., Zhao, K., et al. (). Optimization strategies of fruit detection to overcome the challenge of unstructured background in field orchard environment: A review. Precis. Agric. 24, –. doi: 10./s-023--9
Crossref Full Text | Google Scholar
Tang, Z., Xu, L., Li, H., Chen, M., Shi, X., Zhou, L., et al. (). YOLOC-tiny: a generalized lightweight real-time detection model for multiripeness fruits of large non-green-ripe citrus in unstructured environments. Front. Plant Sci. 15. doi: 10./fpls..
PubMed Abstract | Crossref Full Text | Google Scholar
Tzutalin, D. (). LabelImg. GitHub repository 6, 4. Available online at: https://github.com/tzutalin/labelImg.
Google Scholar
Vrochidou, E., Tsakalidou, V., Kalathas, I., Gkrimpizis, T., Pachidis, T., and Kaburlasos, V. (). An overview of end efectors in agricultural robotic harvesting systems. Agriculture 12, . doi: 10./agriculture
Crossref Full Text | Google Scholar
Wang, C., Bochkovskiy, A., and Liao, H. (). “YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. Piscataway, New Jersey, USA. –.
Google Scholar
Wang, J., Gao, K., Jiang, H., and Zhou, H. (). Method for detecting dragon fruit based on improved lightweight convolutional neural network. Nongye Gongcheng Xuebao/ Trans. Chin. Soc Agric. Eng. 36, 218–225. doi: 10./j.issn.-..20.026
Crossref Full Text | Google Scholar
Wang, C., Yeh, I., and Mark, L. (). “Yolov9: Learning what you want to learn using programmable gradient information,” in European conference on computer vision. 1–21, XXXH. Y. doi: 10./arXiv..
Crossref Full Text | Google Scholar
Wei, J., Ni, L., Luo, L., Chen, M., You, M., Sun, Y., et al. (). GFS-YOLO11: A maturity detection model for multi-variety tomato. Agronomy 14, . doi: 10./agronomy
Crossref Full Text | Google Scholar
Wu, M., Lin, H., Shi, X., Zhu, S., and Zheng, B. (). MTS-YOLO: A multi-task lightweight and efficient model for tomato fruit bunch maturity and stem detection. Horticulturae 10, . doi: 10./horticulturae
Crossref Full Text | Google Scholar
Wu, H., Mo, X., Wen, S., Wu, K., Ye, Y., Wang, Y., et al. (). DNE-YOLO: A method for apple fruit detection in Diverse Natural Environments. J. King Saud University-Computer Inf. Sci. 36, . doi: 10./j.jksuci..
Crossref Full Text | Google Scholar
Yang, J., Ni, J., Li, Y., Wen, J., and Chen, D. (). The intelligent path planning system of agricultural robot via reinforcement learning. Sensors 22, . doi: 10./s
PubMed Abstract | Crossref Full Text | Google Scholar
Yang, Y., Su, L., Zong, A., Tao, W., Xu, X., Chai, Y., et al (). A New Kiwi Fruit Detection Algorithm Based on an Improved Lightweight Network. Agriculture 14 (10), . doi: 10./agriculture
Crossref Full Text | Google Scholar
Zhang, J., Kang, N., Qu, Q., Zhou, L., and Zhang, H. (). Automatic fruit picking technology: a comprehensive review of research advances. Artificial Intelligence Review 57 (3), 54. doi: 10./s-023--2
Crossref Full Text | Google Scholar
Zhao, P., Zhou, W., and Na, L. (). High-precision object detection network for automated pear picking. Sci. Rep. 14, . doi: 10./s-024--6
PubMed Abstract | Crossref Full Text | Google Scholar
Zheng, Z., Wang, P., Liu, W., Li, J., Ye, R., and Ren, D. (). “Distance-IoU loss: Faster and better learning for bounding box regression,” in Proceedings of the AAAI conference on artificial intelligence, Vol. 34. –. Menlo Park, California, USA. doi: 10./aaai.v34i07.
Crossref Full Text | Google Scholar
During the past 70 years, jujube research has greatly advanced. A historic leap has been achieved from the initial stage of focusing on summarizing production experiences to a new era of innovative research throughout the whole industry chain covering breeding, cultivation, pest management, fruit storage, transportation, and processing; from morphological research to comprehensive research at the levels of the plant, organ, tissue, cell, molecule, and gene; and from the research of a few individuals to national and international research collaborations.
Genome sequencing is the golden key to unlocking the genetic code of life. Genome sequencing-based multiomics analysis combining transcriptome, proteome, metabolome, and phenotypic data has taken horticultural research to a new level.
In , de novo genome sequencing of jujube was accomplished (the first in the Rhamnaceae family)12. A strategy combining WGS sequencing, BAC-to-BAC and WGS-PCR-free library was employed to address the impact of the high complexity of the jujube genome, which has a high heterozygosity of 1.90%, a low GC content of 33.41% and a high density of simple-sequence repeats (SSRs) (378.1 per Mb). The assembly (437.6 Mb) covered 98.6% of the estimated jujube genome (444 Mb), and 32,808 protein-coding genes were predicted (http://jujube.genomics.cn/page/species/index.jsp). Later, another jujube genome and the chloroplast genome sequencing of four Ziziphus species were published15,16.
Using an interspecific population between Z. jujuba and Z. acidojujuba, a high-density molecular (SNP) genetic map was constructed by restriction site-associated DNA sequencing17. The joint map across the 12 linkage groups (the same as the basic chromosome number in jujube) spanned .22 cM, with a mean marker distance of 0.32 cM. Combining these genetic linkage groups with the assembled genome sequences, pseudomolecules for each of the 12 chromosomes of jujube were constructed. A total of 23,996 genes (73% of the total annotated genes) were allocated on the 12 pseudochromosomes. Self-alignment of the jujube genome sequences based on the 23,996 gene models identified 943 paralogous gene groups, indicating that the jujube genome may have undergone frequent intrachromosomal fusions and segment duplication during its evolutionary history. A gene block located in the region from 9.20 to 14.68 Mb of pseudochromosome 1 is highly conserved and contains many genes related to sugar metabolism and stress tolerance. An evolutionary divergence analysis of jujube, pear and Prunus mume found that the 4DTv (fourfold synonymous third-codon transversion) rate of jujube peaked at only 0.50, suggesting that no recent whole-genome duplication had occurred in jujube12.
Genome-wide screening of SSR markers and reference genes for RT-qPCR analysis. The density of SSRs in the jujube genome reaches 378 SSRs per Mb, which is approximately two times the density in peach and apple12. Using the genome data, 511 pairs of SSR primers showing high polymorphism were screened and applied for the identification of large-scale hybrid progeny and a genetic diversity analysis18,19,20. A total of 963 jujube germplasms were analyzed by SSRs to construct the core collection and germplasm resource management database21.
Specific reference genes for RT-qPCR of jujube were selected under a variety of conditions and sourced from different tissues/organs, fruit development stages, and biotic/abiotic stresses, providing more choices for further gene expression analysis and functional studies in jujube22,23,24,25,26.
Multiomics-based analysis for the molecular formation mechanisms of some important traits. Combining comparative genome, transcriptome, and metabolome data, the molecular mechanism underlying the high contents of ascorbic acid (AsA) and sugar in jujube fruit (~100 and 2 times those of apple, respectively) were revealed. The l-galactose pathway is the major route for jujube AsA biosynthesis, and the genes encoding the key enzymes involved in the biosynthesis pathway show high-level expression during fruit development12. Meanwhile, the expansion of the MDHAR gene family contributes to AsA regeneration. Further studies indicated that GLDH and MDHAR are the crucial genes in jujube AsA synthesis and recycling, respectively27. The high level of sugar accumulation in jujube fruit is due to the expansion and high-level expression of genes involved in sugar metabolism and transport. In addition, the distinct trait of the ‘bearing shoot falling in winter’ is related to the ethylene and abscisic acid (ABA) pathways12.
The genome sequencing of a drying jujube cultivar ‘Junzao’ and the resequencing of some cultivated and wild jujubes identified the selective sweep regions involved in acid and sugar metabolism and provided insights into an important domestication pattern in fruit taste15. Studies have also proven that sugar transport plays a significant role in sugar accumulation15,28 and jujube fruits have characteristics of nonclimacteric fruits15,29. The ethylene and ABA pathways are involved in regulating jujube fruit ripening30,31.
Some gene families involved in phytoplasma and cold stress and flower and fruit development were identified and analyzed at the genome level22,23,24,32,33,34. A series of gene families and metabolisms responsive to phytoplasma infection were studied, showing that photosynthetic, carbohydrate, and energy metabolism play crucial roles during phytoplasma infection33 and that the MAPK-WRKY pathway is responsive to phytoplasma stress23,24,32.
As one of the longest-cultivated fruit trees in the world, jujube has abundant germplasm resources after undergoing long periods of natural evolution and artificial selection. From the s to s, the first nationwide jujube germplasm investigation was carried out; a total of 700 jujube cultivars and 30 wild sour jujube genotypes were determined through synonym and homonym identification and recorded in the book China Fruit Tree Records-Jujube1.
Based on the nationwide germplasm investigation, the National Chinese Jujube Repository was constructed in Taigu, Shanxi Province, by the Ministry of Agriculture and Rural Affairs of the People’s Republic of China. To date, a total of ~930 jujube genotypes have been preserved, accounting for at least 90% of the total jujube genotypes in the world. Another germplasm repository funded by the State Forestry and Grassland Bureau of China was constructed in Cangxian, Hebei Province, where ~640 germplasm accessions, including some excellent variants of the local leading cultivars, such as ‘Jinsixiaozao’, ‘Wuhezao’, and ‘Dongzao’, are maintained. In addition, Hebei Agricultural University collected and preserved ~200 accessions of sour jujube germplasm and some other species of the Ziziphus genus, such as Z. mauritiana Lam., Z. spina-christi Willd, and Z. nummularia Burm. f.
To date, ~700 jujube cultivars and 100 sour jujube genotypes have been evaluated. The evaluations have covered morphological, agronomical, cytological, palynological, nutritional, and reproductive biological traits, as well as biotic resistance and abiotic tolerance2,35,36,37. A number of excellent accessions have been identified, including triploid genotypes such as ‘Zanhuangdazao’38, ‘Pingguozao’39, ‘Jinglinyihaozao’, ‘Shanxitedage’, ‘Hengshuibianzhizao’, ‘Zhenhuluzao’40; the mixoploid genotype ‘Dongzao 2’41; the male-sterile germplasms ‘JMS1’ and ‘JMS2’42; the self-fruitless and self-sterile germplasms ‘Huizao’, ‘Jinsixiaozao 39’, ‘Yuanlingzao’, and ‘Xiangzao’43; the tortuous-branch ‘Dongzao’44; the seedless germplasm ‘Wuhexiaozao’; genotypes that are highly resistant to witches’ broom disease such as ‘Xingguang’45,46; and genotypes that are rich in functional nutrients47,48,49,50,51,52,53,54.
Three representative books on jujube germplasm have been published. The first is China Fruit Tree Records-Jujube, mentioned above. The second is Germplasm Resources of Chinese Jujube10, which recorded accessions of jujube (Z. jujuba Mill.), sour jujube (Z. acidojujuba Liu et Cheng) and Indian jujube (Z. mauritiana L.). The third is The Illustrated Germplasm Resources of Jujube55, in which pictures and research data for 250 cultivars (57 table jujubes, 80 for dehydration, 82 for both dehydration and fresh eating, 17 for processing, and 14 ornamental varieties) were obtained from the National Chinese Jujube Repository (Taigu, Shanxi). In addition, other books such as Descriptors and Data Standards for Jujube Germplasm56, Test Guidelines for Distinctness, Uniformity and Stability—jujube57, Technical Regulators for the Identification of Jujube Cultivars-SSR Marker Method provide references and technical standards for character selection, data collection, testing, and identification in jujube germplasm studies.
The established germplasm platforms include the following:
(1) Internet Information System for Perennial and Asexual Crop Germplasm Resources (http://www.ziyuanpu.net.cn/), which includes the data observed for many years at the National Jujube Repository in Taigu, Shanxi, China;
(2) Chinese Crop Germplasm Resources Information System—jujube (http://www.cgris.net/query/croplist.php), where germplasm checking and analysis can be performed; and
(3) Internet Information System for Jujube Germplasms (http://www.ziziphus.net/zzzy), which provided information observed from local areas about 700 germplasm accessions recorded in China Fruit Tree Records—Jujube.
In addition, the International Cultivar Registration Center for the Ziziphus genus was established in at the Research Center of Chinese Jujube, Hebei Agricultural University under the authorization of the International Society of Horticultural Sciences.
In the book ‘Qi Min Yao Shu’, published YA, it is clearly recorded that ancient Chinese people used to select the trees with the best tasting fruits of wild sour jujube (Z acidojujuba Cheng et Liu—Z. jujuba Mill. var. spinosa Hu) and cultivate them1,2,3,4. Qu et al. proposed that jujube evolved from wild sour jujube1 based on a systematic study of ancient books, ecological distributions, karyotyping, palynology, and isoenzymes as well as of the transitional types between jujube and sour jujube1. A further cluster analysis of isoenzymes, palynology, and DNA showed that some jujube cultivars and sour jujube genotypes were usually clustered together instead of being totally separated58,59,60, indicating that there are probably several evolutionary pathways from sour jujube to jujube. This result was later confirmed by SSR and cpSSR analysis61,62.
As to the original cultivation center of jujube, outside of China, Iran, and Japan had been regarded as candidates by some scholars outside China. However, the earliest records and evidence from Iran and Japan could only be traced back to YA, when Sino–Japan exchanges became popular and Zhang Qian was sent on a diplomatic mission to west Asia and European countries during the Han Dynasty. According to the author’s investigation in Iran, all the jujube trees that were hundreds of years old were located at key sites on the ancient Silk Road, and Iranian scholars report that the jujube was introduced from China. In the Book of Songs, published YA, it is clearly mentioned that jujubes were already widely cultivated in China. Additionally, according to the unearthed carbonized fruits, jujube was cultivated and utilized in China years ago. It was reported that jujube originated from the Yellow River valley between Shaanxi and Shanxi provinces, China, after studying ancient documents, fossils and modern distributions as well as the transitional genotypes of jujube and sour jujube1,5.
Based on the classical taxonomic system, jujube is listed in the Rhamnaceae family, Rhamnales order63. However, the Rhamnaceae family was moved to the Rosales order in the APG III Angiosperm Phylogeny Group Classification based on chloroplast DNA sequencing (Angiosperm Phylogeny Group III, ), which was further confirmed by phylogenic studies on the genome-sequenced plant species using 390 shared genes64.
The classification system for the Ziziphus genus was proposed based on field investigations, textual research on specimens and historical documents65. The genus Ziziphus was grouped into two sections based on geographical distribution, bearing shoot persistence, and leaf hairiness. They are Section Ziziphus Cheng et Liu and Section Perdurans Cheng et Liu. The latter was further divided into Ser. Cymosiflora Cheng et Liu and Ser. Thyrsiflora Cheng et Liu mainly according to the inflorescence type. Most species belong to Sect. Perdurans; only jujube and sour jujube originating from China and Z. lotus L., native to the Mediterranean, belong to Section Ziziphus due to their deciduous bearing shoots.
Regarding the taxonomic relationship between jujube and sour jujube (Z acidojujuba Cheng et Liu—Z. spinosa Hu), four viewpoints have been reported66, i.e., that they are the same species, that jujube is a variety of sour jujube, that sour jujube is a variety of jujube, or that they are two different species. Regarding the obvious differences in distribution, morphology, usage, and historical knowledge of jujube and sour jujube in China, Liu et al. proposed that they could be treated as two different species, Z. jujuba Mill. and Z. acidojujuba Liu et Cheng67,68. However, jujubes and sour jujubes are very closely related, with cross-compatibility and transitional types between them.
The subdivisions of Z. jujuba Mill. and Z. acidojujuba Liu et Cheng were also proposed to consider them as two independent species67. Under Z. jujuba Mill., five forms have been reported, i.e., f. tortuosa Cheng et Liu (Z. jujuba Mill. var. tortuosa Hort., Z. jujuba Mill. cv. Tortuosa), f. lageniformis (Nakai) Kitag. (Z. sativa Gatern. var. lageniformis Nakai, Z. jujuba Mill. var. lageniformis Hort.), f. carnosicalycis (Wang) Cheng et Liu (Z. jujuba Mill. var. carnosicalycis Wang), f. allochroa Cheng et Liu, f. heteroformis Cheng et Liu (Z. jujuba Mill. cv. heteroformis Hort., Z. jujuba Mill. var. quinequeflora Hort.), and f. apyrena Cheng et Liu (Z. jujuba Mill. var. anucleatus Y. G. Chen). Within Z. acidojujuba Liu et Cheng, three forms were confirmed, namely, f. granulata Cheng et Liu, f. trachysperma Cheng et Liu, and f. infecunda Cheng et Liu.
There appear to be at least 2, 16, and 6 scientific names for the jujube genus, the jujube and the sour jujube, respectively, due to different classification viewpoints and poor academic exchange in the past. Based on a systematic study of the historical taxonomic literature, Liu et al. affirmed Ziziphus (rather than Zizyphus), Z. jujuba Mill. and Ziziphus acidojujuba Cheng et Liu as the proper scientific names for the jujube genus, the jujube and the sour jujube, respectively10,65,66,67,68,69.
Jujube breeding has a long history. However, until the end of the 20th century, the breeding techniques for Chinese jujube had been mainly focused on selection from seedlings, bud mutants, and local germplasm, even though some new techniques had been incorporated into selection breeding, such as marker-assisted identification and standardized techniques. After entering the 21st century, great progress has been made in polyploidy and cross-breeding in jujube. Genetic engineering has also made some advances, but these applications are not yet fully developed.
Breeding objectives should take into consideration the characteristics of jujube trees, the demands of all related parties, the breeding trends in fruit trees and breeding practices in jujube70. The overall objective should be to meet the various needs of farmers, processers, marketers and consumers. The specific objectives should include outstanding resistance to biotic and abiotic stresses, dwarfing, low branching ability, thornlessness, early bearing, high and stable yields, high quality, stonelessness, high nutrient levels, various ripening times, ease of transport and storage, and multiple-use cultivars. The major requirements to meet farmers’ needs are reducing inputs, increasing output and accelerating economic returns. Good quality, stonelessness, high nutrient levels, and varied ripening times can satisfy consumers, whose demands have become increasingly critical and diversified. Objectives such as tolerance to transportation and long storage life, varied ripening times, and multiple-use cultivars are marketers’ preferences. A revolutionary novel cultivar could simultaneously satisfy the diverse demands of farmers, processers, marketers, and consumers.
Given the limitations of traditional selection breeding for obtaining breakthrough cultivars and the extreme difficulty of cross-breeding in jujube, polyploid breeding seems to be a promising prospect. Consumers usually prefer large fruits, but increasing fruit size by applying more fertilizer and plant regulators may result in poor fruit quality. As a result, polyploidy induction has become an ideal breeding approach for obtaining new cultivars with high-quality, large fruits.
Four generations of polyploid induction techniques using colchicine as the main mutagen have been developed for jujube, i.e., in vivo apical bud induction71, in vitro apical or lateral bud induction72, in vitro callus/embryo induction73 and in vivo callus induction74. The fourth-generation technique is to induce polyploidy by treating in vivo calluses induced on branch cuts with colchicine, which could eliminate the severe mixoploidy formation from traditional polyploid breeding and directly produce pure polyploid shoots (Fig. 3). The pure polyploid shoot could form flowers and even set fruits for further evaluation in the same year as the polyploid induction. Consequently, the pure polyploid creation and evaluation period can be shortened by 3–5 years. Until now, a total of 25 triploid, tetraploid, and octoploid strains have been created, among which three excellent tetraploid strains have been released as new cultivars by the Hebei Forestry Cultivar Examination and Approval Committee75,76,77. The novel field technique for homogeneous polyploidy induction has been successfully applied in sour jujube78 and in elm (data not shown).
Auto-tetraploids differ greatly from their diploid counterparts in their morphology, cytology, and nutrient content79. Comparing the tetraploid jujube cultivar ‘Riguang’ and its diploid counterpart ‘Dongzao’, ‘Riguang’ has wider and darker green leaves, higher chlorophyll content, a higher photosynthesis rate, larger stomata, larger pollen and flowers, wider and larger fruits, and an earlier maturation time but is also less vigorous and less cold hardy79. The contents of vitamin C, cAMP, soluble sugars, titratable acids, sucrose, glucose, and fructose were significantly higher in tetraploid ‘Riguang’ fruits than in diploid ‘Dongzao’ fruits79. It was discovered that an autotetraploid sour jujube had higher tolerance to salinity than the diploid, and its preliminary molecular mechanism was illustrated80,81. In addition, triploids of ‘Jinsixiaozao’, ‘Yuanlingzao’, and ‘Changhongzao’ have been created by endosperm culture, but no new cultivar has been released82. Anther/pollen culture has also been practiced in jujube, and some plants have been generated from pollen73,83.
Cross-breeding, the most powerful breeding method for fruit trees, has not been successfully utilized in jujube. This is a result of several key obstacles, including the extreme difficulty of emasculation of small flowers (~5 mm in diameter), the low fruit-setting rate (only ~1%) and the high embryo abortion rate. The rate of obtaining hybrids in jujube is usually <0.01% by the traditional crossing approach. Since no new gene fusions or multiple trait combinations are available in autopolyploidization and it is difficult to make large breakthroughs via selection breeding, there is no substitute for cross-breeding, and its advantages are also incomparable.
In the last 10 years, a high-efficiency hybrid breeding technology system combining male-sterile germplasm, embryo rescue, net control hybridization, and molecular identification was established. Hybrid plant production was increased by 100 times using the new system, and a large number of hybrid progeny were obtained from 19 cross combinations, of which a number of superior lines were selected70.
Three emasculation-free methods have been developed based on the discovery of two typical male-sterile germplasms and a group of self-fruitless/self-sterile germplasms that can replace male sterility42,43, which effectively overcame the key obstacle to artificial emasculation in jujube. Method 1: Hybrids are produced by using a male-sterile variety as the female parent and a variety with high pollen viability and compatibility as the male parent20,84. Method 2: A self-fruitless or self-sterile variety is chosen as the female parent, and a variety with high pollen viability and compatibility is chosen as the male parent. In the two cases mentioned above, all the offspring from self-fruitless or self-sterile parents can be directly regarded as authentic hybrids. Method 3: is a universal technology free of emasculation, i.e., covering the parents with nets to keep away unexpected pollen donors, pollinating by bees inside the nets, and identifying the hybrids with molecular markers84. This method can produce cross and reciprocal-cross hybrids at the same time85.
The problem of hybrids not being obtained due to heavy early embryo abortion was solved by embryo rescue based on the understanding of the mechanism of embryo abortion and the factors affecting very young embryo culture. These factors included the culture media, inoculation methods, removal or maintenance of the seed coat, combination of growth regulators, and concentrations of lactalbumin hydrolysate, activated carbon, and sucrose86,87,88. The seedling rate of young embryos <30 days after flowering (before the abortion peak) was increased from 3.7 to 40%89. The key techniques included peeling off the seed coat, culturing young embryos together with their endosperm and reverse inoculating young embryos with the chalazal end pointed down on the medium.
In recent years, the interploidy hybridization of jujube and the interspecific hybridization between jujube and sour jujube were successfully carried out20,84,85. With the breakthroughs in cross-breeding technology, the acquisition of a progeny population and the establishment of genetic maps, genetic research into important traits was also carried out.
Several high-density genetic linkage maps have been constructed based on segregation populations of ‘JMS2’ × ‘Xing16’, ‘Dongzao’ × ‘Linyilizao’, ‘Dongzao’ × ‘Jinsi 4’, and ‘Dongzao’ × ‘Yinshanhong’17,90,91,92. The male sterility and kernelless traits are controlled by homozygous recessive genes, and some QTL loci of quantitative traits have also been identified92,93,94,95.
In the past 30 years, a total of ~200 new cultivars with large fruit, good fruit quality, high resistance to diseases, and varying uses and maturity times were released through polyploid breeding or selection from seedlings, bud mutants, and local germplasms2. Among them, the tetraploids ‘Chenguang’, ‘Hongguang’, ‘Riguang’, and ‘Zhuguang’, bred by Hebei Agricultural University, have been awarded new plant variety rights by the National Forestry and Grassland Bureau of China. The fruits of the tetraploids were 30–50% larger in size, 4–7 days earlier to mature and better tasting than diploid fruits.
A large number of local cultivars, such as ‘Zanhuangdazao’, ‘Linyilizao’, ‘Dongzao’, ‘Qiyuexian’, ‘Junzao’, ‘Huizao’, etc. have been excavated and utilized, which has greatly promoted the rapid development of the jujube industry. Recently, some excellent new cultivars for fresh eating, such as ‘Jinsi 4’, ‘Yueguang’, ‘Zaohongmi’, ‘Zaocumi’, and ‘Zaoqiuhong’, and some for dehydration, such as ‘Yuangling 2’, ‘Shuguang’, ‘Zanshuo’, ‘Yushuai’, and ‘Linhuang 1’, with larger fruit, higher quality and higher resistance to fruit diseases than traditional cultivars, have become the dominant cultivars. These dominant cultivars have replaced the traditional cultivars, which has greatly improved the cultivar structure in China.
A set of new propagation approaches has been developed on the basis of traditional sucker division. Among them, stimulating sucker propagation by cutting off the roots at the periphery of the vertical projection of the canopy and separating fasciculate suckers can increase the reproductive coefficient, while gathering and nurturing the suckers in a nursery can greatly improve the quality of the root system. Jujube hardwood cuttings are quite difficult to propagate, presenting a rooting rate of usually <30%, and have rarely been used in commercial production. However, green shoot cuttings (semilignified primary extension shoots, secondary shoots, and bearing shoots) root much more easily and have a high reproductive coefficient. The rooting rate can reach as high as 95%96, but its cost and technical requirements are relatively higher than those of other propagation techniques.
Judging from old grafted jujube trees, grafting has been used for at least years. To meet the needs of large-scale development, grafting propagation with sour jujube as the rootstock has been widely used since the late s. In particular, the use of sour jujube seeds, rather than pits, to obtain rootstocks has become the mainstream method; using this method, the seedlings grow faster and more uniformly than in rootstocks obtained from pits2. The use of Paliurus hemsleyanus Rehd. as rootstock can also play a role in the prevention of witches’ broom in jujube in southern China97. Jujube has also been successfully grafted onto Indian jujube (Z. mauritiana Lam.) in subtropical and tropical regions.
The tissue culture of jujube began in . In vitro plantlets were obtained from the stem segments of root suckers in . After , research on jujube tissue culture increased rapidly, and tissue culture with stem tips, stem segments, leaves, anthers, embryos, and cotyledons as explants were all successful83,86,87,99,100. However, propagation via tissue culture has not been used on a large scale in jujube in China because of the high technical requirements, high cost, and late fruiting of micropropagated plants.
Research on jujube cultivation technology has a long history that includes fruitful achievements and has played an important role in promoting the jujube industry. The biological characteristics of jujube are basically understood1. To date, cultivation technology systems have been established for the leading cultivars in their main growing areas with their own characteristics. High-density planting and protected cultivation systems have also been applied commercially after the beginning of the 21st century.
Since the s, the biological characteristics of jujube have been studied comprehensively, and a complete theoretical system had been formed by the s1,101. It was revealed that jujube has very strong resistance to abiotic stress, including drought, barren soils, and saline and alkali conditions. It has unique branch and bud characteristics, i.e., usually only the primary shoots can extend, the dormant buds have a very long life, the secondary shoots die back naturally each, the mother-bearing shoots can only extend by ~1 mm per year, and the bearing shoots fall off in the fall. Its flower bud differentiation and fruit set habit are also very distinct, with a short flower bud differentiation time (10 days), a 2-month flowering season, and a low fruit set of only ~1%.
Since the s, various cultivation techniques focusing on high yield were developed for the leading cultivars and main production areas, which increased production by over 50%, increased the high-quality fruit rate by over 30%, and reduced pesticide use by over 50%102,103. At the same time, traditional sparse planting systems with large crowns (row spacing and plant height above 5 m) and intercropping jujubes with cereal crops (row spacing ≥12 m) were replaced by dense dwarf planting (2 m × 3 m) and monoculture orchards. After the Asian Olympic Games in Beijing in , dense cultivation was developed for dwarf fresh jujube, and even superdense plantings (grass orchards) with densities of up to 15,000 plants per hectare were established. After entering the 21st century, in the desert of southern Xinjiang Province, China, a novel cultivation model for high early yields and high fruit quality was established. This method is characterized by the direct sowing of the rootstock seeds (sour jujube) in orchards followed by the in situ grafting of the target cultivar. Starting with a superhigh density (0.5 m × 1.0 m), the density is gradually decreased to 1.0–1.5 m × 4.0 m. This new model obtains good yields (5–8 t/ha) in the year of grafting (Fig. 4) and maintains the high yield at over 15 t/ha 3–5 years later, which is 3–5 years earlier than this yield could be achieved in a traditional orchard104.
After entering the 21st century, protected cultivation techniques for fresh jujube have developed gradually. Plastic house and Chinese solar greenhouse cultivation (a plastic house with thick back wall as a thermal mass) have been successful in North China and have formed large-scale production regions of over 10,000 ha in Dali County in Shaanxi Province and Linyi County in Shanxi Province in China. In this case, the maturity period can be advanced by 1–4 months105,106. These techniques effectively solve the problem of a short supply period for open field cultivation and can increases revenues by 3–5 times. Solar greenhouses along hillsides facing the sun in the Taihang Mountains of Hebei Province have the advantages of lower investment costs, better sunlight, and much better heat retaining properties than a traditional greenhouse by using the mountain as the back wall of the greenhouses107. The cultivation of fresh jujubes in plastic shelters has resulted in great success in rainy southern China108; this method can greatly reduce the fruit cracking caused by rain at the maturity stage from 70 to <10%.
For fresh eating jujubes, it was found that promoting the lignification of bearing shoots through extremely heavy pruning may accelerate fruiting, increase yield and result in larger fruit109. This technique has become a common practice in fresh jujube production. On the other hand, the lignification of bearing shoots changes the deciduous habit of bearing shoots and increases the pruning labor costs.
According to a comprehensive investigation, more than 100 pests and diseases have been observed in jujube1. Only approximately ten of them cause severe yield and quality losses, such as peach fruit moth, jujube inchworm, Ancylis sativa Liu, Tetranychus viennensis Zacher, jujube rust, jujube witches’ broom, and fruit cracking. After the beginning of the 21st century, outbreaks of pests and diseases, including Lygocoris lucorum, Euzophera batangensis, Ceratitis capitata, jujube flies, fruit cracking, and fruit shrinking, have become increasingly severe102,110,111,112,113,114.
High-efficiency management systems for the main diseases and pests in jujube mentioned above have been established in China3,46,103. However, high-efficiency, low-cost practical management systems for fruit cracking and fruit shrinking disease have not been developed; these conditions have a great influence on jujube yield and quality. In addition, due to the economic decline of the comparative benefits of jujube, jujube witches’ broom is becoming serious again in orchards due to poor management.
Jujube is difficult to keep fresh. Under normal room temperature and humidity conditions, fresh jujube loses moisture quickly and loses crispness within 3–5 days. It ferments readily and lose firmness when it is tightly sealed in conventional plastic bags under cold storage. A postharvest physiological study revealed the main factors influencing fruit preservation in jujube115. At present, there are two opposing viewpoints about the respiratory type of jujube fruit, i.e., climacteric or nonclimacteric, even for the same cultivar (‘Dongzao’)116,117,118.
Many reports on fresh fruit preservation techniques in jujube have been published. The practical techniques include cold storage, controlled atmosphere storage, decompression storage, and controlled freezing-point storage119,120,121,122. Qu et al. reported that ‘Shanxixiaozao’ could be stored for 70 days at 0 ± 1 °C and 60% RH119. Chen et al. indicated that jujube was sensitive to CO2 and that fruit browning occurred quickly under the conditions of 10% CO. Zhang et al. found that the respiration rate, ethylene release, and cell membrane permeability of ‘Dongzao’ stored at 0 ± 0.5 °C were significantly inhibited compared with those of jujube stored at room temperature121.
Hypobaric storage can delay ripening and aging and inhibit the fermentation of jujube fruit by providing low-temperature, low-oxygen storage conditions123. Wang et al. proved that the best storage conditions for fresh jujube were at a temperature of −1 to −2 °C, relative humidity of 95%, 2% O2 and 0% CO. Chen et al. showed that the percentage of healthy fruit and the edible rate of ‘Dongzao’ jujube stored at −2 °C for 100 days were 7.4% and 20.2% higher than those of jujube stored at 0 °C125. Fu found that controlled freezing-point storage was better than normal cold storage in terms of delaying ripening and aging126. Currently, under the optimal storage conditions, half-red fresh jujube can be preserved for 2–3 months or even more than 4 months.
However, in addition to the storage conditions, the duration of fresh jujube storage and the percentage of fruit losses are also influenced by the preharvest cultivation technology and directly by the pathogen load on the fruit in the orchard14,100,113,127,128.
In the past 20 years, postharvest treatment technology for Chinese jujube has made great progress. Fruit drying technology has gradually changed from traditional natural drying under the sun to dehydration in a drying room or by drying machine129. It takes ~1 month for natural drying or air drying to be complete, while the time required can be reduced to 1 day or even less with artificial drying. Compared with those under natural drying, the preservation rates for vitamin C, total sugar, sucrose, fructose, glucose, soluble protein, and other nutrients under artificial drying are significantly improved130,131,132.
In recent years, equipment for jujube fruit sorting or integrated cleaning and sorting has been developed and widely applied in the jujube industry in China133,134,135. The combination of artificial drying, mechanical cleaning, and automatic sorting can significantly improve the appearance of commercial fruit, increase production efficiency and economic benefits, and reduce losses after harvest.
To perform nutritional analysis and intensive processing on jujube, efficient extraction and determination methods were established and optimized for 23 kinds of nutrient components (7 vitamins, 3 triterpenic acids, 8 amino acids, aromatics, cAMP, polysaccharides, flavonoids, and pigments) in jujube54,136,137,138,139,140,141,142. Converting dehydroascorbic acid to AsA (vitamin C) resolved the bottleneck problem in the determination of dehydroascorbic acid, and accurate determination methods for the two kinds of vitamin C in jujube were finalized141,143. The nutrient components in different organs, different fruit developmental stages, and different varieties were systematically analyzed53,54,144,145,146,147. Jujube leaves are rich in leucine, vitamin B6, carotene, betulinic acid, and ursolic acid, the flowers are rich in vitamin B1 and leucine, and mature fruits are rich in cAMP, functional sugars, vitamin B, triterpenic acid, proline, and some other important functional components in addition to the well-known carbohydrates and vitamin C.
Jujube is an important traditional herb and tonic. Comprehensive studies have shown that the most advantageous nutritional features of jujube fruit include its contents of soluble sugars (2–3 times the levels in other fruits), vitamin C (100 times the level in other fruits), cAMP ( times the level in other fruits), vitamin B, triterpenoid acid, proline, polysaccharide, flavonoids, iron, potassium, calcium, and zinc. Therefore, jujube has broad prospects as a material for the development of healthy foods with high nutrition value.
There are many traditional processed jujube products, such as candied jujube, smoked jujube, stoneless sugared jujube, jujube liquor, liquor-saturated jujube, jujube jam, jujube paste, and so on103,148,149. In the last 30 years, various new products, such as jujube juice, jujube powder, jujube slices, jujube tea, jujube beer, jujube essence, and jujube pigment, have been developed149.
The company is the world’s best yulu pear for export(ms,pt,ja) supplier. We are your one-stop shop for all needs. Our staff are highly-specialized and will help you find the product you need.
Previous: Rectangular Opening Woven Wire Screen | Red Star
Next: Učinkovitost Proizvajalca samonosnih filtrov: Ste pripravljeni na revolucijo?
If you are interested in sending in a Guest Blogger Submission,welcome to write for us!
All Comments ( 0 )