Non-official implement of Paper：MnasNet: Platform-Aware Neural Architecture Search for Mobile
Non-official implement of Paper：ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design
- Python3
- PyTorch 0.4.1
- tensorboardX (optional)
- torchnet
- pretrainedmodels (optional)
We just test four models in ImageNet-1K, both train set and val set are scaled to 256(minimal side), only use Mirror and RandomResizeCrop as training data augmentation, during validation, we use center crop to get 224x224 patch.
CPU Info: Intel(R) Xeon(R) CPU E5-2650 v3 @ 2.30GHz
| Models | validation(Top-1) | validation(Top-5) | CPU Cost(ms) |
|---|---|---|---|
| MnasNet | 64.91 | 86.28 | ~300 |
| ShuffleNetV2 x1 | 61.83 | 83.99 | ~100 |
Maybe the implement of these network have some different from origin method, we can not achieve the best performance as said in the paper.