Skip to main content
. 2022 Sep 28;13:993244. doi: 10.3389/fpls.2022.993244

TABLE 4.

Comparison of parameter quantity, GFLOPs, inference, inference speed, and GPU resource occupancy of different models.

Methods Parameter quantity (M) GFLOPs Inference (Min) Inference speed (ms) GPU resource occupancy (G)
YOLOv5s 13.38 15.8 370.5 7.5 1.70
YOLOv5m 39.77 47.9 396.2 11.6 1.80
YOLOv5l 87.90 107.6 415.6 17.3 2.10
YOLOv5x 164.36 204.0 479.9 29.0 2.40
Improved YOLOv5s 28.81 31.6 372.5 14.7 2.42
Faster R-CNN 41.30 278.2 755.3 227.7 7.87