Migrations
/
pytorch-image-models
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Add MixNet Small and Large PyTorch native weights (no same padding)

Browse Source
pull/23/head
Ross Wightman 5 years ago
parent 15e66863b1
commit 4fe2da558c
2 changed files with 29 additions and 34 deletions
Show Stats Download Patch File Download Diff File

51
README.md
Unescape View File

@ -69,35 +69,28 @@ A CSV file containing an ImageNet-1K validation results summary for all included
### Self-trained Weights ### Self-trained Weights
I've leveraged the training scripts in this repository to train a few of the models with missing weights to good levels of performance. These numbers are all for 224x224 training and validation image sizing with the usual 87.5% validation crop. I've leveraged the training scripts in this repository to train a few of the models with missing weights to good levels of performance. These numbers are all for 224x224 training and validation image sizing with the usual 87.5% validation crop.
#### @ 224x224 |Model | Prec@1 (Err) | Prec@5 (Err) | Param # | Image Scaling | Image Size |
|Model | Prec@1 (Err) | Prec@5 (Err) | Param # | Image Scaling | |---|---|---|---|---|---|
|---|---|---|---|---| | efficientnet_b2 | 79.760 (20.240) | 94.714 (5.286) | 9.11M | bicubic | 260 |
| resnext50d_32x4d | 79.674 (20.326) | 94.868 (5.132) | 25.1M | bicubic | | resnext50d_32x4d | 79.674 (20.326) | 94.868 (5.132) | 25.1M | bicubic | 224 |
| resnext50_32x4d | 78.512 (21.488) | 94.042 (5.958) | 25M | bicubic | | mixnet_l | 78.976 (21.024 | 94.184 (5.816) | 7.33M | bicubic | 224 |
| resnet50 | 78.470 (21.530) | 94.266 (5.734) | 25.6M | bicubic | | efficientnet_b1 | 78.692 (21.308) | 94.086 (5.914) | 7.79M | bicubic | 240 |
| mixnet_m | 77.256 (22.744) | 93.418 (6.582) | 5.01M | bicubic | | resnext50_32x4d | 78.512 (21.488) | 94.042 (5.958) | 25M | bicubic | 224 |
| seresnext26_32x4d | 77.104 (22.896) | 93.316 (6.684) | 16.8M | bicubic | | resnet50 | 78.470 (21.530) | 94.266 (5.734) | 25.6M | bicubic | 224 |
| efficientnet_b0 | 76.912 (23.088) | 93.210 (6.790) | 5.29M | bicubic | | mixnet_m | 77.256 (22.744) | 93.418 (6.582) | 5.01M | bicubic | 224 |
| resnet26d | 76.68 (23.32) | 93.166 (6.834) | 16M | bicubic | | seresnext26_32x4d | 77.104 (22.896) | 93.316 (6.684) | 16.8M | bicubic | 224 |
| mobilenetv3_100 | 75.634 (24.366) | 92.708 (7.292) | 5.5M | bicubic | | efficientnet_b0 | 76.912 (23.088) | 93.210 (6.790) | 5.29M | bicubic | 224 |
| mnasnet_a1 | 75.448 (24.552) | 92.604 (7.396) | 3.89M | bicubic | | resnet26d | 76.68 (23.32) | 93.166 (6.834) | 16M | bicubic | 224 |
| resnet26 | 75.292 (24.708) | 92.57 (7.43) | 16M | bicubic | | mixnet_s | 75.988 (24.012) | 92.794 (7.206) | 4.13M | bicubic | 224 |
| fbnetc_100 | 75.124 (24.876) | 92.386 (7.614) | 5.6M | bilinear | | mobilenetv3_100 | 75.634 (24.366) | 92.708 (7.292) | 5.5M | bicubic | 224 |
| resnet34 | 75.110 (24.890) | 92.284 (7.716) | 22M | bilinear | | mnasnet_a1 | 75.448 (24.552) | 92.604 (7.396) | 3.89M | bicubic | 224 |
| seresnet34 | 74.808 (25.192) | 92.124 (7.876) | 22M | bilinear | | resnet26 | 75.292 (24.708) | 92.57 (7.43) | 16M | bicubic | 224 |
| mnasnet_b1 | 74.658 (25.342) | 92.114 (7.886) | 4.38M | bicubic | | fbnetc_100 | 75.124 (24.876) | 92.386 (7.614) | 5.6M | bilinear 224 |
| spnasnet_100 | 74.084 (25.916) | 91.818 (8.182) | 4.42M | bilinear | | resnet34 | 75.110 (24.890) | 92.284 (7.716) | 22M | bilinear | 224 |
| seresnet18 | 71.742 (28.258) | 90.334 (9.666) | 11.8M | bicubic | | seresnet34 | 74.808 (25.192) | 92.124 (7.876) | 22M | bilinear | 224 |
| mnasnet_b1 | 74.658 (25.342) | 92.114 (7.886) | 4.38M | bicubic | 224 |
#### @ 240x240 | spnasnet_100 | 74.084 (25.916) | 91.818 (8.182) | 4.42M | bilinear | 224 |
|Model | Prec@1 (Err) | Prec@5 (Err) | Param # | Image Scaling | | seresnet18 | 71.742 (28.258) | 90.334 (9.666) | 11.8M | bicubic | 224 |
|---|---|---|---|---|
| efficientnet_b1 | 78.692 (21.308) | 94.086 (5.914) | 7.79M | bicubic |
#### @ 260x260
|Model | Prec@1 (Err) | Prec@5 (Err) | Param # | Image Scaling |
|---|---|---|---|---|
| efficientnet_b2 | 79.760 (20.240) | 94.714 (5.286) | 9.11M | bicubic |
### Ported Weights ### Ported Weights

12
timm/models/gen_efficientnet.py
Unescape View File

@ -112,10 +112,12 @@ default_cfgs = {
'tf_efficientnet_b7': _cfg( 'tf_efficientnet_b7': _cfg(
url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/tf_efficientnet_b7_aa-076e3472.pth', url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/tf_efficientnet_b7_aa-076e3472.pth',
input_size=(3, 600, 600), pool_size=(19, 19), crop_pct=0.949), input_size=(3, 600, 600), pool_size=(19, 19), crop_pct=0.949),
'mixnet_s': _cfg(url=''), 'mixnet_s': _cfg(
url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/mixnet_s-a907afbc.pth'),
'mixnet_m': _cfg( 'mixnet_m': _cfg(
url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/mixnet_m-4647fc68.pth'), url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/mixnet_m-4647fc68.pth'),
'mixnet_l': _cfg(url=''), 'mixnet_l': _cfg(
url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/mixnet_l-5a9a2ed8.pth'),
'tf_mixnet_s': _cfg( 'tf_mixnet_s': _cfg(
url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/tf_mixnet_s-89d3354b.pth'), url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/tf_mixnet_s-89d3354b.pth'),
'tf_mixnet_m': _cfg( 'tf_mixnet_m': _cfg(
@ -1636,12 +1638,12 @@ def tf_efficientnet_b7(pretrained=False, num_classes=1000, in_chans=3, **kwargs)
def mixnet_s(pretrained=False, num_classes=1000, in_chans=3, **kwargs): def mixnet_s(pretrained=False, num_classes=1000, in_chans=3, **kwargs):
"""Creates a MixNet Small model. """Creates a MixNet Small model.
""" """
default_cfg = default_cfgs['mixnet_m'] default_cfg = default_cfgs['mixnet_s']
model = _gen_mixnet_s( model = _gen_mixnet_s(
channel_multiplier=1.0, num_classes=num_classes, in_chans=in_chans, **kwargs) channel_multiplier=1.0, num_classes=num_classes, in_chans=in_chans, **kwargs)
model.default_cfg = default_cfg model.default_cfg = default_cfg
#if pretrained: if pretrained:
# load_pretrained(model, default_cfg, num_classes, in_chans) load_pretrained(model, default_cfg, num_classes, in_chans)
return model return model

Write Preview
Loading…
Cancel
Save