4 ms·

AGQA: A Benchmark for Compositional, Spatio-Temporal Reasoning

• p1esk 5y ago
  Actual dataset is here: https://cs.stanford.edu/people/ranjaykrishna/agqa/ https://cs.stanford.edu/people/ranjaykrishna/agqa/ It's 192M questions about this dataset (9k videos): https://prior.allenai.org/projects/charades https://prior.allenai.org/projects/charades
• infogulch 5y ago
  I'm interested in how they structured the spatio-temporal scene graph annotations. E.g. what is the general schema?