Asparagus Detection Method Based on Attention Mechanism

Asparagus detection in greenhouse environments frequently encounters challenges due to complex backgrounds, leading to diminished accuracy. We proposes YOLOv10s-iRMB, an improved asparagus detection algorithm. Our method integrates an Inverted Residual Mobile Block (iRMB) into the YOLOv10s architecture. This design leverages depth-wise convolution to significantly reduce parameters. This improves feature extraction and overall accuracy. Residual connections within the module preserve information from earlier layers, promoting robust and stable training. Experimental results show that YOLOv10s-iRMB achieves a detection accuracy of 88.77%, marking a 9.69% improvement than YOLOv10s. Additionally, [email protected]:0.95 has also been significantly improved.