The final, formatted version of the article will be published soon.
ORIGINAL RESEARCH article
Front. Plant Sci.
Sec. Technical Advances in Plant Science
Volume 15 - 2024 |
doi: 10.3389/fpls.2024.1455687
This article is part of the Research Topic Emerging Sustainable and Green Technologies for Improving Agricultural Production View all 10 articles
YOLOv5-POS: Research on cabbage Pose Prediction Method Based on multi-task Perception Technology
Provisionally accepted- School of Information Engineering, Huzhou University, Huzhou, China
Accurate and rapid identification of cabbage posture is essential for mechanical harvesters to minimize damage to cabbage heads during the harvesting process. This study introduces YOLOv5-POS, an innovative cabbage posture prediction approach. Building on the YOLOv5s backbone, this method enhances detection and segmentation capabilities for cabbage heads and roots by incorporating C-RepGFPN to replace the traditional Neck layer, optimizing feature extraction and upsampling strategies, and refining the C-Seg segmentation head. Additionally, a cabbage root growth prediction model based on Bé zier curves is proposed, using the geometric moment method for key point identification and the anti-gravity stem-seeking principle to determine root-head junctions. It performs precision root growth curve fitting and prediction, effectively overcoming the challenge posed by the outer leaves completely enclosing the cabbage root stem. Validated on a multi-variety cabbage image dataset, YOLOv5-POS achieves an F1 score of 98.8% for head and root detection, with an average instance segmentation accuracy of 93.5%. The average absolute error and average relative error in cabbage pose recognition are recorded as 1.38° and 2.32°, respectively. The prediction model achieves an average accuracy of 98%, with cabbage posture recognition requiring only 28 milliseconds. This method effectively supports real-time cabbage harvesting.
Keywords: Multi-task perception network, Cabbage harvest, YOLOv5-POS, Bezier curve, posture recognition
Received: 27 Jun 2024; Accepted: 30 Aug 2024.
Copyright: © 2024 Shen, Shao, Cheng, Zhou and Yao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence:
Cheng Zhou, School of Information Engineering, Huzhou University, Huzhou, China
Lili Yao, School of Information Engineering, Huzhou University, Huzhou, China
Disclaimer: 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.