VGG-PyTorch

Overview

This repository contains an op-for-op PyTorch reimplementation of Very Deep Convolutional Networks for Large-Scale Image Recognition.

Table of contents

• VGG-PyTorch
  • Overview
  • Table of contents
  • Download weights
  • Download datasets
  • How Test and Train
    • Test
    • Train model
    • Resume train model
  • Result
  • Contributing
  • Credit
    • Very Deep Convolutional Networks for Large-Scale Image Recognition

Download weights

• Google Driver
• Baidu Driver

Download datasets

Contains MNIST, CIFAR10&CIFAR100, TinyImageNet_200, MiniImageNet_1K, ImageNet_1K, Caltech101&Caltech256 and more etc.

• Google Driver
• Baidu Driver

Please refer to README.md in the data directory for the method of making a dataset.

How Test and Train

Both training and testing only need to modify the test_config.py file.

Test

Modify test_config.py file.

• line 17: model_arch_name change to vgg11.
• line 31: model_num_classes change to 1000.
• line 24: mode change to ./data/ImageNet_1K/ILSVRC2012_img_val.
• line 37: model_weights_path change to ./results/pretrained_models/VGG11-ImageNet_1K-64f6524f.pth.tar.

python3 test.py

Train model

Modify train_config.py file.

• line 18: model_arch_name change to vgg11.
• line 19: model_num_classes change to 1000.
• line 25: mode change to ./data/ImageNet_1K/ILSVRC2012_img_train.
• line 26: mode change to ./data/ImageNet_1K/ILSVRC2012_img_val.
• line 37: pretrained_model_weights_path change to ./results/pretrained_models/VGG11-ImageNet_1K-64f6524f.pth.tar.

python3 train.py

Resume train model

Modify train_config.py file.

• line 18: model_arch_name change to vgg11.
• line 19: model_num_classes change to 1000.
• line 25: mode change to ./data/ImageNet_1K/ILSVRC2012_img_train.
• line 26: mode change to ./data/ImageNet_1K/ILSVRC2012_img_val.
• line 40: resume_model_weights_path change to ./samples/VGG11-ImageNet_1K/epoch_xxx.pth.tar.

python3 train.py

Result

Source of original paper results: https://arxiv.org/pdf/1409.1556v6.pdf)

In the following table, the top-x error value in () indicates the result of the project, and - indicates no test.

Model	Dataset	Top-1 error (val)	Top-5 error (val)
VGG11	ImageNet_1K	29.6%(30.9%)	10.4%(11.3%)
VGG11_BN	ImageNet_1K	-(29.6%)	-(10.2%)
VGG13	ImageNet_1K	28.7%(30.1%)	9.9%(10.8%)
VGG13_BN	ImageNet_1K	-(28.4%)	-(9.6%)
VGG16	ImageNet_1K	27.0%(28.4%)	8.8%(9.6%)
VGG16_BN	ImageNet_1K	-(26.6%)	-(8.5%)
VGG19	ImageNet_1K	27.3%(27.6%)	9.0%(9.1%)
VGG19_BN	ImageNet_1K	-(25.7%)	-(8.1%)

# Download `VGG11-ImageNet_1K-64f6524f.pth.tar` weights to `./results/pretrained_models`
# More detail see `README.md<Download weights>`
python3 ./inference.py 

Input:

Output:

Build VGG11 model successfully.
Load VGG11 model weights `/VGG-PyTorch/results/pretrained_models/VGG11-ImageNet_1K-64f6524f.pth.tar` successfully.
tench, Tinca tinca                                                          (74.97%)
barracouta, snoek                                                           (23.09%)
gar, garfish, garpike, billfish, Lepisosteus osseus                         (0.81%)
reel                                                                        (0.45%)
armadillo                                                                   (0.25%)

Contributing

If you find a bug, create a GitHub issue, or even better, submit a pull request. Similarly, if you have questions, simply post them as GitHub issues.

I look forward to seeing what the community does with these models!

Credit

Very Deep Convolutional Networks for Large-Scale Image Recognition

Karen Simonyan, Andrew Zisserman

Abstract

In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3×3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16–19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

[Paper]

@article{simonyan2014very,
  title={Very deep convolutional networks for large-scale image recognition},
  author={Simonyan, Karen and Zisserman, Andrew},
  journal={arXiv preprint arXiv:1409.1556},
  year={2014}
}