Overview

Memory-Augmented Video Semantic Network for learning to:

• Encode semantic descriptors of video into an embedding
• Predict genres and IMDB ratings

Paper

Conference of Neural Information Processing Systems (NIPS 2017): Visually-Grounded Interaction and Language Workshop : PDF

Datasets

Movie clip dataset

The dataset attached here is a subset of samples that were used for training our model. We use several such video segments from the same movie to train our model.

• Full annotations (.csv)
• Aggregated annotations (.csv)

Contact

Have a question or comment? Please contact us at pralav [at] mit [dot] edu