MusicVideoClipRetrival

This is a project for my Masters degree, for the Multimodal master degree course

In order to run run the application the following steps are necessary:

1. Include the data folder in the same directory (90 most popular songs in Spotify
2. Install the libraries in the requirements.txt
3. To set up the image data for the Resnet-50 model run the frames_set_up.ipynb to create the frames for each music video clip.
4. To set up the text data for the Bert-uncased model run the lyrics_set_up.ipynb to create both text files that include the lyrics and csv files that include vital information for the content of the music video clips.
5. Run the resnet_model.ipynb, that will create the trained torch model.
6. Run the bert_base_model to get the results of the textual model.