Table 6.
Inference performance and resource utilization comparison of models.
| Model | Number of parameters/model size (MB) | Inference time (ms) | Frames per second (FPS) | Floating point operations per second (FLOPS) |
|---|---|---|---|---|
| VGG16 | 15,668,070/59.77 MB | 68 | 14.5 | 15.347 |
| Resnet-50 | 25,724,838/98.13 MB | 72 | 13.5 | 3.864 |
| Efficient Net-B0 | 4,098,249/15.63 MB | 69 | 14.5 | 0.385 |
| DenseNet-121 | 7,076,454/26.99 MB | 80 | 12.3 | 2.839 |
| MobileNet-V2 | 3,267,814/12.47 MB | 62 | 16.5 | 0.567 |
| Inception-V3 | 23,939,910/91.32 MB | 75 | 12.5 | 5.784 |
| Inception-ResnetV2 | 54,395,142/207.50 MB | 70 | 13.5 | 6.551 |
| Xception | 20,939,342/79.88 MB | 71 | 13.5 | 3.601 |
| Proposed CNN-SEEIB | 3,340,194/12.74 MB | 64 | 16.5 | 2.121 |