Table 4. Training results.
The YOLOv5 model has the highest number of layers, whereas the ResNet64 model contains the most parameters.
Model | Layer | Parameters | GFLOPs | Top1 | Time (h) |
---|---|---|---|---|---|
ResNet64 | 175 | 25,557,032 | 0.0511 | 0.930 | 0.174 |
YOLOv5 | 214 | 7,030,417 | 16.0 | 0.882 | 0.304 |
YOLOv8 | 99 | 27,192,88 | 4.4 | 0.937 | 0.139 |