No description, website, or topics provided.
Switch branches/tags
Nothing to show
Clone or download
Latest commit 5a484b6 Mar 14, 2018
Permalink
Failed to load latest commit information.
utils added files Mar 14, 2018
LICENSE added license Mar 14, 2018
README.md Create README.md Mar 14, 2018
eval.py added files Mar 14, 2018
inference.py added files Mar 14, 2018
train.py added files Mar 14, 2018

README.md

NeuralNetwork-Viterbi: A Framework for Weakly Supervised Video Learning

Code for the paper NeuralNetwork-Viterbi: A Framework for Weakly Supervised Video Learning

Prepraration:

  • download the data from https://uni-bonn.sciebo.de/s/vVexqxzKFc6lYJx
  • extract it so that you have the data folder in the same directory as train.py
  • create a results directory in the same directory where you also find train.py: mkdir results

Requirements: Python2.7 with the libraries numpy and pytorch

Training:

Run ./train.py

Inference:

Run ./infernece.py Note: adjust the variable n_threads in inference.py to your needs.

Evaluation:

In the inference step, recognition files are written to the results directory. The frame-level ground truth is available in data/groundTruth. Run ./eval.py --recog_dir=results --ground_truth_dir=data/groundTruth to evaluate the frame accuracy of the trained model.

Remarks:

We provide a python/pytorch implementation for easy usage. In the paper, we used an internal C++ implementation, so results can be slightly different. Running the provided setup on split1 of Breakfast should lead to roughly 42% frame accuracy.

If you use the code, please cite

A. Richard, H. Kuehne, A. Iqbal, J. Gall:
NeuralNetwork-Viterbi: A Framework for Weakly Supervised Video Learning
in IEEE Int. Conf. on Computer Vision and Pattern Recognition, 2018