Fig. 1. Network structure of YOLOv5

Fig. 2. Structure of SPP

Fig. 3. Structure of attention mechanism. (a) SE model structure; (b) spatial attention model structure in CBAM

Fig. 4. Improved YOLOv5_FI network structure

Fig. 5. Pre-processing of fingerprint image. (a) Original image; (b) background segmentation; (c) local ridge direction; (d) ridge enhancement; (e) binarization

Fig. 6. Image annotation

Fig. 7. Dataset amplification effect

Fig. 8. Network structure. (a) YOLOv5s_8; (b) YOLOv5s_16; (c) YOLOv5s_32

Fig. 9. Network structure of YOLOv5s_B

Fig. 10. Network structure of YOLOv5s_C

Fig. 11. YOLOv5s_ Attention network and dataset related information. (a) Network structure of YOLOv5s_Attention; (b) distribution of target locations in dataset; (c) proportion of target size of dataset

Fig. 12. Training of YOLOv5s_FI. (a) Obj_loss change curve; (b) cls_loss change curve

Fig. 13. Comparison of detection results. (a) YOLOv5s_FI; (b) YOLOv5s