Out of Sight but Not Out of Mind: Hybrid Memory for Dynamic Video World Models
Paper • 2603.25716 • Published • 157
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
HM-World is a dataset for hybrid memory in dynamic video world models. It provides video sequences, foreground masks, camera pose annotations, character pose annotations, event timestamps, and text captions for each sample.
Merge the split archives and extract the dataset with:
cat HM-World_* | tar -xzvf -
HM-World/
├── sample1/
│ ├── cond.mp4
│ ├── tgt.mp4
│ ├── cond_mask.mp4
│ ├── tgt_mask.mp4
│ ├── camera.json
│ ├── character.json
│ └── check.json
├── sample2/
│ └── ...
├── ...
└── caption.txt
cond.mp4: condition video.tgt.mp4: target video.cond_mask.mp4: foreground mask video for the condition video.tgt_mask.mp4: foreground mask video for the target video.camera.json: camera pose information.character.json: character pose information.check.json: timestamps that record when the subjects enters or leaves the frame.caption.txt: captions for all samples in the dataset.