Improved multi-classes kiwifruit detection in orchard to avoid collisions during robotic picking

Suo, Rui; Gao, Feng; Zhou, Zhongxian; Fu, Longsheng; Song, Zhenzhen; Dhupia, Jaspreet Singh; Li, Rui; Cui, Yongjie

doi:10.1016/j.compag.2021.106052

Cited by 70 publications

(21 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the trends of the convergence curve, the three models learned the object features well, and all the loss values after stabilization were less than 1, which shows that the models can be used in detection, similar to the literature [33]. We evaluate the training results and compare the AP values of the three detection models on banana bunches and stalks, and the mAP values of the entire model for all detection classes [33]. The two calculation formulas are (1) and ( 2):…”

Section: Model Evaluationsupporting

confidence: 76%

“…It can be seen from the figure that the loss of YOLO-Banana is higher than the loss of YOLOv4 before stabilization, and the loss after stabilization is between YOLOv4 and YOLO-Banana-l4; although YOLO-Banana-l4 converges early, the decline rate of loss is the slowest, and the loss is higher than that of YOLO-Banana after 1300 iterations. From the trends of the convergence curve, the three models learned the object features well, and all the loss values after stabilization were less than 1, which shows that the models can be used in detection, similar to the literature [33].…”

Section: Model Evaluationsupporting

confidence: 75%

“…EfficientDet trades off the speed and accuracy of the neural network; it was used by [31] to reconstruct the 3D global mapping of the orchard. YOLO series, which will be introduced in the following section, were used for cucumber internode length [32], kiwifruits [33], grapefruits [34], grapes [35], banana bunches [36], and banana bunches and stalks [37]. At the same time, improved networks based on YOLO series, namely MangoYOLO [38], YOLO-Tomato [39], and YOLOMuskmelon [40], were proposed; beyond these, the improved YOLOv3 was applied to detect the banana inflorescence axis [41] and an improved YOLOv5 method was described for apple detection [42].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

YOLO-Banana: A Lightweight Neural Network for Rapid Detection of Banana Bunches and Stalks in the Natural Environment

Yang

et al. 2022

Agronomy

View full text Add to dashboard Cite

The real-time detection of banana bunches and stalks in banana orchards is a key technology in the application of agricultural robots. The complex conditions of the orchard make accurate detection a difficult task, and the light weight of the deep learning network is an application trend. This study proposes and compares two improved YOLOv4 neural network detection models in a banana orchard. One is the YOLO-Banana detection model, which analyzes banana characteristics and network structure to prune the less important network layers; the other is the YOLO-Banana-l4 detection model, which, by adding a YOLO head layer to the pruned network structure, explores the impact of a four-scale prediction structure on the pruning network. The results show that YOLO-Banana and YOLO-Banana-l4 could reduce the network weight and shorten the detection time compared with YOLOv4. Furthermore, YOLO-Banana detection model has the best performance, with good detection accuracy for banana bunches and stalks in the natural environment. The average precision (AP) values of the YOLO-Banana detection model on banana bunches and stalks are 98.4% and 85.98%, and the mean average precision (mAP) of the detection model is 92.19%. The model weight is reduced from 244 to 137 MB, and the detection time is shortened from 44.96 to 35.33 ms. In short, the network is lightweight and has good real-time performance and application prospects in intelligent management and automatic harvesting in the banana orchard.

show abstract

Section: Model Evaluationsupporting

confidence: 76%

Section: Model Evaluationsupporting

confidence: 75%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

YOLO-Banana: A Lightweight Neural Network for Rapid Detection of Banana Bunches and Stalks in the Natural Environment

Yang

et al. 2022

Agronomy

View full text Add to dashboard Cite

show abstract

“…Monhollen et al [7] developed a corn kernel loss rate detection program based on Faster R-CNN, which achieved an average accuracy of 0.90, the additional field tests obtained the accuracy of 0.91. Suo et al [8] used yolov4 to study the transfer of kiwifruit detection, and obtained the highest mAP of 91.9% with an image processing speed of 25.5 ms. Zhang et al [9] proposed a water-meter pointerreading recognition method based on improved yolov4; the detection accuracy of this method reached 98.68%, which indicated that the lightweight algorithm could quickly and accurately identify targets. Li et al [10] proposed a rapid detection model for green pepper based on yolov4-tiny, the average precision is 95.11%, the model size is 30.9 MB, and the frame rate is 89 FPS.…”

Section: Introductionmentioning

confidence: 99%

Design and Experiment of a Broken Corn Kernel Detection Device Based on the Yolov4-Tiny Algorithm

Yao

et al. 2021

Agriculture

View full text Add to dashboard Cite

At present, the wide application of the CNN (convolutional neural network) algorithm has greatly improved the intelligence level of agricultural machinery. Accurate and real-time detection for outdoor conditions is necessary for realizing intelligence and automation of corn harvesting. In view of the problems with existing detection methods for judging the integrity of corn kernels, such as low accuracy, poor reliability, and difficulty in adapting to the complicated and changeable harvesting environment, this paper investigates a broken corn kernel detection device for combine harvesters by using the yolov4-tiny model. Hardware construction is first designed to acquire continuous images and processing of corn kernels without overlap. Based on the images collected, the yolov4-tiny model is then utilized for training recognition of the intact and broken corn kernels samples. Next, a broken corn kernel detection algorithm is developed. Finally, the experiments are carried out to verify the effectiveness of the broken corn kernel detection device. The laboratory results show that the accuracy of the yolov4-tiny model is 93.5% for intact kernels and 93.0% for broken kernels, and the value of precision, recall, and F1 score are 92.8%, 93.5%, and 93.11%, respectively. The field experiment results show that the broken kernel rate obtained by the designed detection device are in good agreement with that obtained by the manually calculated statistic, with differentials at only 0.8%. This study provides a technical reference of a real-time method for detecting a broken corn kernel rate.

show abstract

“…On the other hand, the one-stage recognition speed is relatively fast, so it is more suitable for mobile applications, because it can directly predict the category and location of objects through features extracted from the network. Commonly used algorithms include YOLOv3 (you only look once v3) [20][21][22], YOLOv4 [23,24], SSD [25], Retina-Net [26], etc. Hou et al [27] proposed a ginger shoot identification method based on YOLOv3, but this method only identifies ginger shoots, resulting in a complex process of calculating ginger shoot orientation, and it had a redundant backbone network.…”

Section: Introductionmentioning

confidence: 99%

Ginger Seeding Detection and Shoot Orientation Discrimination Using an Improved YOLOv4-LITE Network

Fang

et al. 2021

Agronomy

View full text Add to dashboard Cite

A consistent orientation of ginger shoots when sowing ginger is more conducive to high yields and later harvesting. However, current ginger sowing mainly relies on manual methods, seriously hindering the ginger industry’s development. Existing ginger seeders still require manual assistance in placing ginger seeds to achieve consistent ginger shoot orientation. To address the problem that existing ginger seeders have difficulty in automating seeding and ensuring consistent ginger shoot orientation, this study applies object detection techniques in deep learning to the detection of ginger and proposes a ginger recognition network based on YOLOv4-LITE, which, first, uses MobileNetv2 as the backbone network of the model and, second, adds coordinate attention to MobileNetv2 and uses Do-Conv convolution to replace part of the traditional convolution. After completing the prediction of ginger and ginger shoots, this paper determines ginger shoot orientation by calculating the relative positions of the largest ginger shoot and the ginger. The mean average precision, Params, and giga Flops of the proposed YOLOv4-LITE in the test set reached 98.73%, 47.99 M, and 8.74, respectively. The experimental results show that YOLOv4-LITE achieved ginger seed detection and ginger shoot orientation calculation, and that it provides a technical guarantee for automated ginger seeding.

show abstract

Improved multi-classes kiwifruit detection in orchard to avoid collisions during robotic picking

Cited by 70 publications

References 32 publications

YOLO-Banana: A Lightweight Neural Network for Rapid Detection of Banana Bunches and Stalks in the Natural Environment

YOLO-Banana: A Lightweight Neural Network for Rapid Detection of Banana Bunches and Stalks in the Natural Environment

Design and Experiment of a Broken Corn Kernel Detection Device Based on the Yolov4-Tiny Algorithm

Ginger Seeding Detection and Shoot Orientation Discrimination Using an Improved YOLOv4-LITE Network

Contact Info

Product

Resources

About