Model params 22 MB
Estimates for a single full pass of model at input size 300 x 300:
- Memory required for features: 37 MB
- Flops: 1 GFLOPs
Estimates are given below of the burden of computing the conv17_2_relu features in the network for different input sizes using a batch size of 128:
| input size | feature size | feature memory | flops |
|---|---|---|---|
| 150 x 150 | 1 x 1 x 128 | 1 GB | 39 GFLOPs |
| 300 x 300 | 1 x 1 x 128 | 4 GB | 146 GFLOPs |
| 450 x 450 | 1 x 1 x 128 | 10 GB | 336 GFLOPs |
| 600 x 600 | 2 x 2 x 128 | 17 GB | 574 GFLOPs |
| 750 x 750 | 2 x 2 x 128 | 27 GB | 890 GFLOPs |
| 900 x 900 | 2 x 2 x 128 | 39 GB | 1 TFLOPs |
A rough outline of where in the network memory is allocated to parameters and features and where the greatest computational cost lies is shown below. The x-axis does not show labels (it becomes hard to read for networks containing hundreds of layers) - it should be interpreted as depicting increasing depth from left to right. The goal is simply to give some idea of the overall profile of the model:
