• Report.pdf - The report of our project.
• Presentation.pdf - The presentation of our project.

Run the process by executing the following scripts or notebooks,

• To extract features from all 3 modalities execute the function get_features(test=False) from feature extraction/get_features.py.
• To extract the embeddings of the features using the AutoEncoder, execute the notebook representation_learning.ipynb.
• To see the experiments or execute your own, execute the notebook video_queries.ipynb.

To execute the experiment you just specify the YouTube title of the video-clip/query and it is automatically downloaded in the folder ../data/data_test.

Note : To run your own experiment make sure to create the folder ../data/data_test, where the root is the current repository. This folder should be empty every time you add a query.