https://github.com/maicius/nima4images
修改model zoo 中的google nima 模型的pytorch实现，增加对批量图片的测试
https://github.com/maicius/nima4images
Last synced: about 1 year ago
JSON representation
修改model zoo 中的google nima 模型的pytorch实现，增加对批量图片的测试
- Host: GitHub
- URL: https://github.com/maicius/nima4images
- Owner: Maicius
- License: mit
- Created: 2018-11-12T08:07:17.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2018-11-12T08:51:49.000Z (over 7 years ago)
- Last Synced: 2025-04-05T10:51:09.246Z (over 1 year ago)
- Language: Python
- Size: 10.8 MB
- Stars: 6
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# PyTorch NIMA: Neural IMage Assessment
PyTorch implementation of [Neural IMage Assessment](https://arxiv.org/abs/1709.05424) by Hossein Talebi and Peyman Milanfar. You can learn more from [this post at Google Research Blog](https://research.googleblog.com/2017/12/introducing-nima-neural-image-assessment.html).
## Installing
```bash
git clone https://github.com/truskovskiyk/nima.pytorch.git
cd nima.pytorch
virtualenv -p python3.6 env
source ./env/bin/activate
pip install -r requirements/linux_gpu.txt
```
or You can just use ready [Dockerfile](./Dockerfile)
## Dataset
The model was trained on the [AVA (Aesthetic Visual Analysis) dataset](http://refbase.cvc.uab.es/files/MMP2012a.pdf)
You can get it from [here](https://github.com/mtobeiyf/ava_downloader)
Here are some examples of images with theire scores
## Model
Used MobileNetV2 architecture as described in the paper [Inverted Residuals and Linear Bottlenecks: Mobile Networks for Classification, Detection and Segmentation](https://arxiv.org/pdf/1801.04381).
## Pre-train model
You can use this [pretrain-model](https://s3-us-west-1.amazonaws.com/models-nima/pretrain-model.pth) with
```bash
val_emd_loss = 0.079
test_emd_loss = 0.080
```
## Deployment
Deployed model on [heroku](https://www.heroku.com/) URL is https://neural-image-assessment.herokuapp.com/ You can use it for testing in Your own images, but pay attention, that's free service, so it cannot handel too many requests. Here is simple curl command to test deployment models
```bash
curl -X POST -F "file=@123.jpg" https://neural-image-assessment.herokuapp.com/api/get_scores
```
Please use our [swagger](https://neural-image-assessment.herokuapp.com/apidocs) for interactive testing
## Usage
```bash
export PYTHONPATH=.
export PATH_TO_AVA_TXT=/storage/DATA/ava/AVA.txt
export PATH_TO_IMAGES=/storage/DATA/images/
export PATH_TO_CSV=/storage/DATA/ava/
export BATCH_SIZE=16
export NUM_WORKERS=2
export NUM_EPOCH=50
export INIT_LR=0.0001
export EXPERIMENT_DIR_NAME=/storage/experiment_n0001
```
Clean and prepare dataset
```bash
python nima/cli.py prepare_dataset --path_to_ava_txt $PATH_TO_AVA_TXT \
--path_to_save_csv $PATH_TO_CSV \
--path_to_images $PATH_TO_IMAGES
```
Train model
```bash
python nima/cli.py train_model --path_to_save_csv $PATH_TO_CSV \
--path_to_images $PATH_TO_IMAGES \
--batch_size $BATCH_SIZE \
--num_workers $NUM_WORKERS \
--num_epoch $NUM_EPOCH \
--init_lr $INIT_LR \
--experiment_dir_name $EXPERIMENT_DIR_NAME
```
Use tensorboard to tracking training progress
```bash
tensorboard --logdir .
```
Validate model on val and test datasets
```bash
python nima/cli.py validate_model --path_to_model_weight ./pretrain-model.pth \
--path_to_save_csv $PATH_TO_CSV \
--path_to_images $PATH_TO_IMAGES \
--batch_size $BATCH_SIZE \
--num_workers $NUM_EPOCH
```
Get scores for one image
```bash
python nima/cli.py get_image_score --path_to_model_weight ./pretrain-model.pth --path_to_image test_image.jpg
```
## Contributing
Contributing are welcome
## License
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details
## Acknowledgments
* [neural-image-assessment in keras](https://github.com/titu1994/neural-image-assessment)
* [Neural-IMage-Assessment in pytorch](https://github.com/kentsyx/Neural-IMage-Assessment)
* [pytorch-mobilenet-v2](https://github.com/tonylins/pytorch-mobilenet-v2)
* [origin NIMA article](https://arxiv.org/abs/1709.05424)
* [origin MobileNetV2 article](https://arxiv.org/pdf/1801.04381)
* [Post at Google Research Blog](https://research.googleblog.com/2017/12/introducing-nima-neural-image-assessment.html)
* [Heroku: Cloud Application Platform](https://www.heroku.com/)