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] [project] [notebook].

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.

Get Started

• Clone this repo and install dependencies:

git clone https://github.com/materight/RepNet-pytorch
cd RepNet-pytorch
pip install -r requirements.txt

• Download the pre-trained weights from Hugging Face.

Run inference

Simply run:

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):

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: