Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/xiadingZ/video-caption.pytorch
pytorch implementation of video captioning
https://github.com/xiadingZ/video-caption.pytorch
deep-learning pytorch video-captioning
Last synced: 2 months ago
JSON representation
pytorch implementation of video captioning
- Host: GitHub
- URL: https://github.com/xiadingZ/video-caption.pytorch
- Owner: xiadingZ
- License: mit
- Created: 2018-01-07T03:14:44.000Z (about 7 years ago)
- Default Branch: master
- Last Pushed: 2019-08-19T11:25:58.000Z (over 5 years ago)
- Last Synced: 2024-05-23T06:49:41.673Z (8 months ago)
- Topics: deep-learning, pytorch, video-captioning
- Language: Python
- Homepage:
- Size: 96.2 MB
- Stars: 398
- Watchers: 11
- Forks: 128
- Open Issues: 25
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- Awesome-pytorch-list-CNVersion - video-caption-pytorch
README
# pytorch implementation of video captioning
recommend installing pytorch and python packages using Anaconda
## requirements
- cuda
- pytorch 0.4.0
- python3
- ffmpeg (can install using anaconda)
### python packages
- tqdm
- pillow
- pretrainedmodels
- nltk
## Data
MSR-VTT. Test video doesn't have captions, so I spilit train-viedo to train/val/test. Extract and put them in `./data/` directory
- train-video: [download link](https://drive.google.com/file/d/1Qi6Gn_l93SzrvmKQQu-drI90L-x8B0ly/view?usp=sharing)
- test-video: [download link](https://drive.google.com/file/d/10fPbEhD-ENVQihrRvKFvxcMzkDlhvf4Q/view?usp=sharing)
- json info of train-video: [download link](https://drive.google.com/file/d/1LcTtsAvfnHhUfHMiI4YkDgN7lF1-_-m7/view?usp=sharing)
- json info of test-video: [download link](https://drive.google.com/file/d/1Kgra0uMKDQssclNZXRLfbj9UQgBv-1YE/view?usp=sharing)
## Options
all default options are defined in opt.py or corresponding code file, change them for your like.
## Acknowledgements
Some code refers to [ImageCaptioning.pytorch](https://github.com/yunjey/pytorch-tutorial/tree/master/tutorials/03-advanced/image_captioning)
## Usage
### (Optional) c3d features
you can use [video-classification-3d-cnn-pytorch](https://github.com/kenshohara/video-classification-3d-cnn-pytorch) to extract features from video.
### Steps
1. preprocess videos and labels
```bash
python prepro_feats.py --output_dir data/feats/resnet152 --model resnet152 --n_frame_steps 40 --gpu 4,5
python prepro_vocab.py
```
2. Training a model
```bash
python train.py --gpu 0 --epochs 3001 --batch_size 300 --checkpoint_path data/save --feats_dir data/feats/resnet152 --model S2VTAttModel --with_c3d 1 --c3d_feats_dir data/feats/c3d_feats --dim_vid 4096
```
3. test
opt_info.json will be in same directory as saved model.
```bash
python eval.py --recover_opt data/save/opt_info.json --saved_model data/save/model_1000.pth --batch_size 100 --gpu 1
```
## TODO
- lstm
- beam search
- reinforcement learning
- dataparallel (broken in pytorch 0.4)
## Acknowledgements
Some code refers to [ImageCaptioning.pytorch](https://github.com/ruotianluo/ImageCaptioning.pytorch)