AUTHOR=Wang Aichen , Peng Tao , Cao Huadong , Xu Yifei , Wei Xinhua , Cui Bingbo 

TITLE=TIA-YOLOv5: An improved YOLOv5 network for real-time detection of crop and weed in the field

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2022.1091655

DOI=10.3389/fpls.2022.1091655

ISSN=1664-462X

ABSTRACT=The effective control of weeds can ensure the yield of crops. Development of weed and crop detection algorithms provides theoretical support for weed control and becomes an effective tool for the site-specific weed management. For weed and crop object detection tasks in the field, there is often a large difference between the number of weed and crop, resulting in an unbalanced distribution of samples and further posing difficulties for the detection task. In addition, most developed models tend to miss the small weed objects, leading to unsatisfied detection results. To overcome these issues, a pixel-level synthesization data augmentation method was proposed to deal with the problem of unbalanced sample distribution of weed and crop. And an improved YOLOv5 network named the TIA-YOLOv5 was proposed for real-time weed and crop detection, in which a transformer encoder block was added to the backbone to improve the sensitivity of the model to weeds, a channel feature fusion with involution (CFFI) strategy was proposed for channel feature fusion while reducing information loss, and adaptive spatial feature fusion (ASFF) was introduced for feature fusion of different scales in the prediction head. Test results with a publicly available sugarbeet dataset showed that the proposed TIA-YOLOv5 network yielded an F1-scoreweed, APweed and mAP@0.5 of 70.0%, 80.8% and 90.0%, respectively, which was 11.8%, 11.3% and 5.9% higher than the baseline YOLOv5 model. And the detection speed reached 20.8 FPS. When deployed on a Jetson NX (NVIDIA, US) computing platform, the TIA-YOLOv5 model could yield an FPS of about 70 after optimization by the TensorRT SDK provided by NVIDIA, which is very promising for real-time weed and crop detection in the field.