You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
244 lines
7.5 KiB
244 lines
7.5 KiB
# ResNeXt
|
|
|
|
A **ResNeXt** repeats a [building block](https://paperswithcode.com/method/resnext-block) that aggregates a set of transformations with the same topology. Compared to a [ResNet](https://paperswithcode.com/method/resnet), it exposes a new dimension, *cardinality* (the size of the set of transformations) $C$, as an essential factor in addition to the dimensions of depth and width.
|
|
|
|
## How do I use this model on an image?
|
|
To load a pretrained model:
|
|
|
|
```python
|
|
import timm
|
|
model = timm.create_model('resnext101_32x8d', pretrained=True)
|
|
model.eval()
|
|
```
|
|
|
|
To load and preprocess the image:
|
|
```python
|
|
import urllib
|
|
from PIL import Image
|
|
from timm.data import resolve_data_config
|
|
from timm.data.transforms_factory import create_transform
|
|
|
|
config = resolve_data_config({}, model=model)
|
|
transform = create_transform(**config)
|
|
|
|
url, filename = ("https://github.com/pytorch/hub/raw/master/images/dog.jpg", "dog.jpg")
|
|
urllib.request.urlretrieve(url, filename)
|
|
img = Image.open(filename).convert('RGB')
|
|
tensor = transform(img).unsqueeze(0) # transform and add batch dimension
|
|
```
|
|
|
|
To get the model predictions:
|
|
```python
|
|
import torch
|
|
with torch.no_grad():
|
|
out = model(tensor)
|
|
probabilities = torch.nn.functional.softmax(out[0], dim=0)
|
|
print(probabilities.shape)
|
|
# prints: torch.Size([1000])
|
|
```
|
|
|
|
To get the top-5 predictions class names:
|
|
```python
|
|
# Get imagenet class mappings
|
|
url, filename = ("https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt", "imagenet_classes.txt")
|
|
urllib.request.urlretrieve(url, filename)
|
|
with open("imagenet_classes.txt", "r") as f:
|
|
categories = [s.strip() for s in f.readlines()]
|
|
|
|
# Print top categories per image
|
|
top5_prob, top5_catid = torch.topk(probabilities, 5)
|
|
for i in range(top5_prob.size(0)):
|
|
print(categories[top5_catid[i]], top5_prob[i].item())
|
|
# prints class names and probabilities like:
|
|
# [('Samoyed', 0.6425196528434753), ('Pomeranian', 0.04062102362513542), ('keeshond', 0.03186424449086189), ('white wolf', 0.01739676296710968), ('Eskimo dog', 0.011717947199940681)]
|
|
```
|
|
|
|
Replace the model name with the variant you want to use, e.g. `resnext101_32x8d`. You can find the IDs in the model summaries at the top of this page.
|
|
|
|
To extract image features with this model, follow the [timm feature extraction examples](https://rwightman.github.io/pytorch-image-models/feature_extraction/), just change the name of the model you want to use.
|
|
|
|
## How do I finetune this model?
|
|
You can finetune any of the pre-trained models just by changing the classifier (the last layer).
|
|
```python
|
|
model = timm.create_model('resnext101_32x8d', pretrained=True, num_classes=NUM_FINETUNE_CLASSES)
|
|
```
|
|
To finetune on your own dataset, you have to write a training loop or adapt [timm's training
|
|
script](https://github.com/rwightman/pytorch-image-models/blob/master/train.py) to use your dataset.
|
|
|
|
## How do I train this model?
|
|
|
|
You can follow the [timm recipe scripts](https://rwightman.github.io/pytorch-image-models/scripts/) for training a new model afresh.
|
|
|
|
## Citation
|
|
|
|
```BibTeX
|
|
@article{DBLP:journals/corr/XieGDTH16,
|
|
author = {Saining Xie and
|
|
Ross B. Girshick and
|
|
Piotr Doll{\'{a}}r and
|
|
Zhuowen Tu and
|
|
Kaiming He},
|
|
title = {Aggregated Residual Transformations for Deep Neural Networks},
|
|
journal = {CoRR},
|
|
volume = {abs/1611.05431},
|
|
year = {2016},
|
|
url = {http://arxiv.org/abs/1611.05431},
|
|
archivePrefix = {arXiv},
|
|
eprint = {1611.05431},
|
|
timestamp = {Mon, 13 Aug 2018 16:45:58 +0200},
|
|
biburl = {https://dblp.org/rec/journals/corr/XieGDTH16.bib},
|
|
bibsource = {dblp computer science bibliography, https://dblp.org}
|
|
}
|
|
```
|
|
|
|
<!--
|
|
Type: model-index
|
|
Collections:
|
|
- Name: ResNeXt
|
|
Paper:
|
|
Title: Aggregated Residual Transformations for Deep Neural Networks
|
|
URL: https://paperswithcode.com/paper/aggregated-residual-transformations-for-deep
|
|
Models:
|
|
- Name: resnext101_32x8d
|
|
In Collection: ResNeXt
|
|
Metadata:
|
|
FLOPs: 21180417024
|
|
Parameters: 88790000
|
|
File Size: 356082095
|
|
Architecture:
|
|
- 1x1 Convolution
|
|
- Batch Normalization
|
|
- Convolution
|
|
- Global Average Pooling
|
|
- Grouped Convolution
|
|
- Max Pooling
|
|
- ReLU
|
|
- ResNeXt Block
|
|
- Residual Connection
|
|
- Softmax
|
|
Tasks:
|
|
- Image Classification
|
|
Training Data:
|
|
- ImageNet
|
|
ID: resnext101_32x8d
|
|
Crop Pct: '0.875'
|
|
Image Size: '224'
|
|
Interpolation: bilinear
|
|
Code: https://github.com/rwightman/pytorch-image-models/blob/b9843f954b0457af2db4f9dea41a8538f51f5d78/timm/models/resnet.py#L877
|
|
Weights: https://download.pytorch.org/models/resnext101_32x8d-8ba56ff5.pth
|
|
Results:
|
|
- Task: Image Classification
|
|
Dataset: ImageNet
|
|
Metrics:
|
|
Top 1 Accuracy: 79.3%
|
|
Top 5 Accuracy: 94.53%
|
|
- Name: resnext50_32x4d
|
|
In Collection: ResNeXt
|
|
Metadata:
|
|
FLOPs: 5472648192
|
|
Parameters: 25030000
|
|
File Size: 100435887
|
|
Architecture:
|
|
- 1x1 Convolution
|
|
- Batch Normalization
|
|
- Convolution
|
|
- Global Average Pooling
|
|
- Grouped Convolution
|
|
- Max Pooling
|
|
- ReLU
|
|
- ResNeXt Block
|
|
- Residual Connection
|
|
- Softmax
|
|
Tasks:
|
|
- Image Classification
|
|
Training Data:
|
|
- ImageNet
|
|
ID: resnext50_32x4d
|
|
Crop Pct: '0.875'
|
|
Image Size: '224'
|
|
Interpolation: bicubic
|
|
Code: https://github.com/rwightman/pytorch-image-models/blob/b9843f954b0457af2db4f9dea41a8538f51f5d78/timm/models/resnet.py#L851
|
|
Weights: https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnext50_32x4d_ra-d733960d.pth
|
|
Results:
|
|
- Task: Image Classification
|
|
Dataset: ImageNet
|
|
Metrics:
|
|
Top 1 Accuracy: 79.79%
|
|
Top 5 Accuracy: 94.61%
|
|
- Name: resnext50d_32x4d
|
|
In Collection: ResNeXt
|
|
Metadata:
|
|
FLOPs: 5781119488
|
|
Parameters: 25050000
|
|
File Size: 100515304
|
|
Architecture:
|
|
- 1x1 Convolution
|
|
- Batch Normalization
|
|
- Convolution
|
|
- Global Average Pooling
|
|
- Grouped Convolution
|
|
- Max Pooling
|
|
- ReLU
|
|
- ResNeXt Block
|
|
- Residual Connection
|
|
- Softmax
|
|
Tasks:
|
|
- Image Classification
|
|
Training Data:
|
|
- ImageNet
|
|
ID: resnext50d_32x4d
|
|
Crop Pct: '0.875'
|
|
Image Size: '224'
|
|
Interpolation: bicubic
|
|
Code: https://github.com/rwightman/pytorch-image-models/blob/b9843f954b0457af2db4f9dea41a8538f51f5d78/timm/models/resnet.py#L869
|
|
Weights: https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnext50d_32x4d-103e99f8.pth
|
|
Results:
|
|
- Task: Image Classification
|
|
Dataset: ImageNet
|
|
Metrics:
|
|
Top 1 Accuracy: 79.67%
|
|
Top 5 Accuracy: 94.87%
|
|
- Name: tv_resnext50_32x4d
|
|
In Collection: ResNeXt
|
|
Metadata:
|
|
FLOPs: 5472648192
|
|
Parameters: 25030000
|
|
File Size: 100441675
|
|
Architecture:
|
|
- 1x1 Convolution
|
|
- Batch Normalization
|
|
- Convolution
|
|
- Global Average Pooling
|
|
- Grouped Convolution
|
|
- Max Pooling
|
|
- ReLU
|
|
- ResNeXt Block
|
|
- Residual Connection
|
|
- Softmax
|
|
Tasks:
|
|
- Image Classification
|
|
Training Techniques:
|
|
- SGD with Momentum
|
|
- Weight Decay
|
|
Training Data:
|
|
- ImageNet
|
|
ID: tv_resnext50_32x4d
|
|
LR: 0.1
|
|
Epochs: 90
|
|
Crop Pct: '0.875'
|
|
LR Gamma: 0.1
|
|
Momentum: 0.9
|
|
Batch Size: 32
|
|
Image Size: '224'
|
|
LR Step Size: 30
|
|
Weight Decay: 0.0001
|
|
Interpolation: bilinear
|
|
Code: https://github.com/rwightman/pytorch-image-models/blob/9a25fdf3ad0414b4d66da443fe60ae0aa14edc84/timm/models/resnet.py#L842
|
|
Weights: https://download.pytorch.org/models/resnext50_32x4d-7cdf4587.pth
|
|
Results:
|
|
- Task: Image Classification
|
|
Dataset: ImageNet
|
|
Metrics:
|
|
Top 1 Accuracy: 77.61%
|
|
Top 5 Accuracy: 93.68%
|
|
--> |