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

Abstract

IntroductionDevelopment 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, we proposed a pixel-level synthesization data augmentation method and a TIA-YOLOv5 network for weed and crop detection in the complex field environment.MethodsThe pixel-level synthesization data augmentation method generated synthetic images by pasting weed pixels into original images. In the TIA-YOLOv5, 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.ResultsTest 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.DiscussionIn this paper, a fast and accurate workflow including a pixel-level synthesization data augmentation method and a TIA-YOLOv5 network was proposed for real-time weed and crop detection in the field. The proposed method improved the detection accuracy and speed, providing very promising detection results.

Weed detection in soybean fields using improved YOLOv7 and evaluating herbicide reduction efficacy

“…In a similar weed detection work, this work is very challenging due to the similarity between crops and weeds. To overcome these problems, Wang A. et al. (2022) proposed a pixel-level integrated data enhancement method and the TIA-YOLOv5 network for weed and crop detection in complex field environments, and their F 1 and mAP were 0.7 and 0.9, respectively.…”

A Hybrid Text Recognition Approach based on Involution Operators and Graph Convolution Neural Networks

“…The IV-IDM intrusion detection model proposed by Han et al (2022) used involution, which was a good solution to the above problems. Wang et al (2022) proposed a channel feature fusion with involution (CFFI) strategy for channel feature fusion. It also reduced information loss and therefore real-time weed and crop detection in the field is achieved.…”

Evaluation of two deep learning‐based approaches for detecting weeds growing in cabbage

“…Previous studies have documented the effectiveness of YOLOv5 38 for weed detection in various crops such as corn (Zea mays L.), pepper (Capsicum annum L.), sugarcane (Saccharum officinarum L.), spinach (Spinacia oleracea L.), 39 and sugarbeet (Beta vulgaris L.). 40 Wang et al 41 examined an improved version of YOLOv5 with an attention mechanism resulting in a weed recognition accuracy of 83% in corn. YOLOv8 is a recent release by Ultralytics in January 2023.…”

“…Previous studies have documented the effectiveness of YOLOv5 38 for weed detection in various crops such as corn ( Zea mays L.), pepper ( Capsicum annum L.), sugarcane ( Saccharum officinarum L.), spinach ( Spinacia oleracea L.), 39 and sugarbeet ( Beta vulgaris L.) 40 . Wang et al 41 .…”