ChinaOpen is a new video dataset targeted at open-world multimodal learning, with raw data gathered from Bilibili, a popular Chinese video-sharing website. The dataset has a large webly annotated training set of videos (associated with user-generated titles and tags) and a smaller manually annotated test set of videos (with manually checked user titles / tags, manually written captions, and manual labels describing what visual objects / actions / scenes shown in the visual content).
1 PAPER • 1 BENCHMARK
Sakuga-42M is a large-scale hand-drawn cartoon video dataset for academic research purposes, it comprises 42 million cartoon keyframes covering various artistic styles, regions, and years, with comprehensive semantic annotations including video-text description pairs, anime tags, content taxonomies, etc. The dataset is intended to support researchers in their exploration of more effective and practical solutions for creating cartoons.
1 PAPER • 2 BENCHMARKS