Improving Blood Cell Detection Accuracy using an Attention-based YOLOv5 Network
Shuangyan Zhu, Yuan Zhou, Yuting Liu, Ming Hu, Daobin Huang · 2023
Accurate quantification of blood cells samples plays a critical role in clinical diagnosis. In this study, we have developed a deep learning network based on YOLOv5 that integrates attention mechanisms into the backbone network for precise detection of blood cells. Extensive experiments were conducted on BCCD dataset to evaluate the effectiveness of our approach. Our experimental results demonstrate the impact of incorporating attention mechanisms into the YOLOv5 network. Notably, our method achieved an impressive precision rate of 89.8%, surpassing the original result by 1.7 percentage points. This improvement signifies the power of attention mechanisms in enhancing the precision of blood cell detection.