Real-Time Grading of Defect Apples Using Semantic Segmentation Combination with a Pruned YOLO V4 Network

Abstract

At present, the apple grading system usually conveys apples by a belt or rollers. This usually leads to low hardness or expensive fruits being bruised, resulting in economic losses. In order to realize real-time detection and classification of high-quality apples, separate fruit trays were designed to convey apples and used to prevent apples from being bruised during image acquisition. A semantic segmentation method based on the BiSeNet V2 deep learning network was proposed to segment the defective parts of defective apples. BiSeNet V2 for apple defect detection obtained a slightly better result in MPA with a value of 99.66%, which was 0.14 and 0.19 percentage points higher than DAnet and Unet, respectively. A model pruning method was used to optimize the structure of the YOLO V4 network. The detection accuracy of defect regions in apple images was further improved by the pruned YOLO V4 network. Then, a surface mapping method between the defect area in apple images and the actual defect area was proposed to accurately calculate the defect area. Finally, apples on separate fruit trays were sorted according to the number and area of defects in the apple images. The experimental results showed that the average accuracy of apple classification was 92.42%, and the F1 score was 94.31. In commercial separate fruit tray grading and sorting machines, it has great application potential.

Detection of early bruises on apples using near‐infrared camera imaging technology combined with adaptive threshold segmentation algorithm

“…In recent years, the rapid development of machine learning, especially deep learning, has provided powerful tools and methods for solving practical problems in various fields. Traditional machine learning methods, such as support vector machines (Li et al, 2022; Su et al, 2022), random forests (Feng et al, 2022), k‐nearest neighbors (Nturambirwe et al, 2021), deep learning methods (Liu et al, 2022), and so forth, such as target detection algorithms (Yao et al, 2021; Yuan et al, 2022), semantic segmentation algorithm (Liang et al, 2022), and so forth, combined with machine vision systems have been widely used in the field of fruit bruise detection and have achieved significant results.…”

Enhancing Shelf Life Prediction of Fresh Pizza with Regression Models and Low Cost Sensors

“…Liang et al. [ 20 ] proposed a separate fruit tray system and a deep learning-based method to detect and classify high-quality apples in real-time. Basak et al.…”

FastQAFPN-YOLOv8s-based method for rapid and lightweight detection of walnut unseparated material

“…Literature [11] proposes a model to improve YOLOv7 citrus lightweight target detection by introducing small target detection, the CBAM attention mechanism for fusion and extraction of multiscale features, and a reduction in the number of model parameters. Literature [12] uses a model pruning method to optimize the structure of the YOLOv4 network, resulting in improved detection accuracy of defective regions in apple images, with a final average classification accuracy of 92.42% and an F1 score of 94.31. Literature [13] proposes a YOLOv7 network and multiple data expansion based oil tea fruit detection method for oil tea fruit detection in complex field scenes, resulting in final mAP, Precision, Recall, F1 scores, and average detection time per image of 96.03%, 94.76%, 95.54%, 95.15%, and 0.025 s, respectively.…”