Based on the YOLOv8 target detection framework,an improved blood cell detection method is proposed.The SIOU loss function is introduced,which can better deal with the skewed shape of the target object in blood cell detection and improve the detection accuracy.The FasterNet Block module is introduced to improve the efficiency and accuracy of feature extraction by opti-mizing the convolution operation and feature fusion strategy.Comparison experiments are conducted on the publicly available blood cell image dataset.The experimental results show that compared with the original model,the improved method achieves significant performance improvement in several evaluation indexes,with an increase of 2.1 percentage point in the mean accuracy(mAP)and 23.2 ms in the detection speed.
关键词
血细胞检测/YOLOv8/SIoU损失函数/FasterBlock模块/深度学习
Key words
blood cell testing/YOLOv8/SIoU loss function/FasterNet block/deep learning
万方数据