Skip to main content
. 2024 Sep 11;14(18):2640. doi: 10.3390/ani14182640

Table 4.

Comparison of different feature extraction backbone networks in head networks.

Methods Precision (%) Recall (%) mAP50 (%) mAP50-95(%) Para (M) GFLOPs
YOLOv8n 85.6 81.6 85.6 65 30.06 8.1
YOLOv8n + CSPS 90.1 84.1 90.3 67.8 29.83 8.1
CSPS + EMA 89.3 84.2 89.6 67.3 29.91 8.2
CSPS + iRMB 85.1 84.6 88.5 67.4 32.79 8.9
CSA 90.3 86.4 86.5 68.5 33.54 8.1