
Smart Agricultural Technology, Год журнала: 2025, Номер unknown, С. 101068 - 101068
Опубликована: Июнь 1, 2025
Язык: Английский
Smart Agricultural Technology, Год журнала: 2025, Номер unknown, С. 101068 - 101068
Опубликована: Июнь 1, 2025
Язык: Английский
Agriculture, Год журнала: 2025, Номер 15(4), С. 372 - 372
Опубликована: Фев. 10, 2025
With the development of intelligent modern agriculture, strawberry harvesting robots play an increasingly important role in precision agriculture. However, existing vision systems face multiple challenges complex farmland environments, including fruit occlusion, difficulties recognizing fruits at varying ripeness levels, and limited real-time processing capabilities. This study proposes a keypoint detection 3D localization method for utilizing depth camera to address these challenges. By introducing Haar Wavelet Downsampling (HWD) module Gold-YOLO neck, proposed achieves significant improvements feature extraction performance. The integration HWD effectively reduces image noise, enhances accuracy, strengthens method’s ability recognize stems. Additionally, incorporating neck structure multi-scale fusion, improving accuracy enabling adapt environments. To further accelerate inference speed enable deployment embedded system, Layer-adaptive sparsity Magnitude-based Pruning (LAMP) technology is employed, significantly reducing redundant parameters thereby enhancing lightweight performance model. Experimental results demonstrate that can accurately identify strawberries different stages exhibits strong robustness under various lighting conditions scenarios, achieving average 97.3% while model 38.2% original model, efficiency localization. provides robust technical support practical application widespread adoption agricultural robots.
Язык: Английский
Процитировано
0Smart Agricultural Technology, Год журнала: 2025, Номер unknown, С. 100948 - 100948
Опубликована: Апрель 1, 2025
Язык: Английский
Процитировано
0Engineering Applications of Artificial Intelligence, Год журнала: 2025, Номер 154, С. 110918 - 110918
Опубликована: Апрель 29, 2025
Язык: Английский
Процитировано
0Опубликована: Май 17, 2025
The development of computer vision and deep learning has promoted agricultural automation. YOLO series algorithms are widely used in fields such as robot fruit picking, but still face challenges occlusion light changes. This study is based on YOLOv5 6.1. C3 module lightweight processed the 5s model to obtain C3-L module. In experiment, was replaced with at positions Backbone, Head Backbone+Head respectively, CBAM CA attention mechanisms were introduced compare performances different models. results show that improved can reduce resource invocation graphics card memory usage during training. stability replacing part good. After adding mechanism, overall accuracy rate increases by 5%. When requirement not high, partially Backbone call hardware resources decrease video 17.4%, which conducive operation mobile hardware. provides a reference for optimization algorithm picking scenarios its transplantation devices microcontrollers.
Язык: Английский
Процитировано
0Engineering Applications of Artificial Intelligence, Год журнала: 2025, Номер 156, С. 111162 - 111162
Опубликована: Май 31, 2025
Язык: Английский
Процитировано
0Smart Agricultural Technology, Год журнала: 2025, Номер unknown, С. 101068 - 101068
Опубликована: Июнь 1, 2025
Язык: Английский
Процитировано
0