MovieNet Movie Understanding Dataset

MovieNet is a dataset for movie understanding that contains 1,100 movies with a large amount of multimodal data, such as trailers, photos, plot descriptions, etc. In addition, MovieNet provides manual annotations of different aspects, including 1.1 million characters with bounding boxes and identity annotations, 42,000 scene boundaries, 2,500 aligned description sentences, 65,000 location and action labels, and 92,000 movie style labels.

The dataset was released by the Chinese University of Hong Kong.