Table 3.
Comparison of pest detection efficiency between different models.
| Models | GFLOPs | Parameters (MB) | Model size (MB) | Inference time (ms) |
|---|---|---|---|---|
| Faster RCNN | 206.8 | 41.24 | 331.2 | 40 |
| Cascade RCNN | 234.6 | 69.0 | 553.3 | 49.5 |
| YOLOv4 | 141.8 | 64.06 | 257.7 | 17.7 |
| YOLOv4-tiny | 16.2 | 5.93 | 22.6 | 4.44 |
| Mobilenetv3-YOLOv4 | 17.0 | 11.43 | 56.9 | 11.2 |
| YOLOv5s | 16.0 | 7.08 | 14.5 | 6.1 |
| AgriPest-YOLO | 16.2 | 7.35 | 15.1 | 8.8 |
GFLOPS indicates Giga Floating-point Operations Per Second; Parameter indicates the total number of parameters of the model; Inference time indicates the inference time on one single image.