Wheat Seed Detection and Counting Method Based on Improved YOLOv8 Model

Abstract

Wheat seed detection has important applications in calculating thousand-grain weight and crop breeding. In order to solve the problems of seed accumulation, adhesion, and occlusion that can lead to low counting accuracy, while ensuring fast detection speed with high accuracy, a wheat seed counting method is proposed to provide technical support for the development of the embedded platform of the seed counter. This study proposes a lightweight real-time wheat seed detection model, YOLOv8-HD, based on YOLOv8. Firstly, we introduce the concept of shared convolutional layers to improve the YOLOv8 detection head, reducing the number of parameters and achieving a lightweight design to improve runtime speed. Secondly, we incorporate the Vision Transformer with a Deformable Attention mechanism into the C2f module of the backbone network to enhance the network's feature extraction capability and improve detection accuracy. The results show that in the stacked scenes with impurities (severe seed adhesion), the YOLOv8-HD model achieves an average detection accuracy (mAP) of 77.6%, which is 9.1% higher than YOLOv8. In all scenes, the YOLOv8-HD model achieves an average detection accuracy (mAP) of 99.3%, which is 16.8% higher than YOLOv8. The memory size of the YOLOv8-HD model is 6.35 MB, approximately 4/5 of YOLOv8. The GFLOPs of YOLOv8-HD decrease by 16%. The inference time of YOLOv8-HD is 2.86 ms (on GPU), which is lower than YOLOv8. Finally, we conducted numerous experiments and the results showed that YOLOv8-HD outperforms other mainstream networks in terms of mAP, speed, and model size. Therefore, our YOLOv8-HD can efficiently detect wheat seeds in various scenarios, providing technical support for the development of seed counting instruments.

Keywords: YOLOv8; attention mechanism; lightweight; wheat seed detection.

Figure 1

Wheat seed collection example images: (a) scattered without impurities; (b) scattered with impurities; (c) clustered without impurities; (d) clustered with impurities; (e) stacked without impurities; (f) stacked with impurities.

Figure 1

Wheat seed collection example images: (a) scattered without impurities; (b) scattered with impurities; (c) clustered without impurities; (d) clustered with impurities; (e) stacked without impurities; (f) stacked with impurities.

Figure 2

Wheat seed data augmented images: (a) original image; (b) augmented image 1; (c) augmented image 2; (d) augmented image 3; (e) augmented image 4; (f) augmented image 5.

Figure 3

Improved YOLOv8 lightweight network model.

Figure 4

YOLOv8 head.

Figure 5

Lightweight YOLOv8 detection head.

Figure 6

Comparison between DAT and other Vision Transformer models: (a) VIT; (b) Swin Transformer; (c) DCN; (d) DAT.

Figure 7

Information flow of the deformable attention mechanism in DAT.

Figure 8

C2f model.

Figure 9

DAT network parameters.

Figure 10

Training loss (above) and validation loss (below) curves of YOLOv8-HD and YOLOv8 models.

Figure 11

Performance curves of YOLOv8-HD and YOLOv8 models.

Figure 12

Comparison between YOLOv8-HD and YOLOv8 heatmaps: (a) YOLOv8 heatmaps, where the red boxes indicate regions with less prominent feature extraction.; (b) YOLOv8-HD heatmaps.

Figure 13

Comparison of FP/FN for YOLOv8-HD and YOLOv8.

Figure 14

Visualization of YOLOv8-HD detection results: (a) dispersed clean; (b) dispersed cluttered; (c) clustered clean; (d) clustered cluttered; (e) clustered clean with cluttered; (f) clustered cluttered.

Figure 14

Visualization of YOLOv8-HD detection results: (a) dispersed clean; (b) dispersed cluttered; (c) clustered clean; (d) clustered cluttered; (e) clustered clean with cluttered; (f) clustered cluttered.

Figure 15

Wheat seed counting results: (a) wheat seed counting results in dispersed scenario; (b) wheat seed counting results in clustered scenario; (c) wheat seed counting results in stacked scenario.