https://github.com/materight/repnet-pytorch
A PyTorch port with pre-trained weights of RepNet, from "Counting Out Time: Class Agnostic Video Repetition Counting in the Wild".
https://github.com/materight/repnet-pytorch
deep-learning pytorch self-supervised-learning
Last synced: 6 months ago
JSON representation
A PyTorch port with pre-trained weights of RepNet, from "Counting Out Time: Class Agnostic Video Repetition Counting in the Wild".
- Host: GitHub
- URL: https://github.com/materight/repnet-pytorch
- Owner: materight
- Created: 2023-02-28T15:28:43.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2024-11-19T12:58:07.000Z (6 months ago)
- Last Synced: 2024-11-19T13:49:58.804Z (6 months ago)
- Topics: deep-learning, pytorch, self-supervised-learning
- Language: Python
- Homepage:
- Size: 6.75 MB
- Stars: 24
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# RepNet PyTorch
A PyTorch port with pre-trained weights of **RepNet**, from *Counting Out Time: Class Agnostic Video Repetition Counting in the Wild* (CVPR 2020) [[paper]](https://arxiv.org/abs/2006.15418) [[project]](https://sites.google.com/view/repnet) [[notebook]](https://colab.research.google.com/github/google-research/google-research/blob/master/repnet/repnet_colab.ipynb#scrollTo=FUg2vSYhmsT0).
This repo provides an implementation of RepNet written in PyTorch and a script to convert the pre-trained TensorFlow weights provided by the authors. The outputs of the two implementations are almost identical, with a small deviation (less than $10^{-6}$ at most) probably caused by the [limited precision of floating point operations](https://pytorch.org/docs/stable/notes/numerical_accuracy.html).
## Get Started
- Clone this repo and install dependencies:
```bash
git clone https://github.com/materight/RepNet-pytorch
cd RepNet-pytorch
pip install -r requirements.txt
```
- Download the pre-trained weights from [Hugging Face](https://huggingface.co/materight/repnet/blob/main/pytorch_weights.pth).
## Run inference
Simply run:
```bash
python run.py --weights [weights_path]
```
The script will download a sample video, run inference on it and save the count visualization. You can also specify a video path as argument (either a local path or a YouTube/HTTP URL):
```bash
python run.py --weights [weights_path] --video_path [video_path]
```
If the model does not produce good results, try to run the script with more stride values using `--strides`.
Example of generated videos showing the repetition count, with the periodicity score and the temporal self-similarity matrix:
