Statements (26)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:model
video transformer |
| gptkbp:author |
Gedas Bertasius
|
| gptkbp:bench |
Diving48
Kinetics-400 Something-Something V2 |
| gptkbp:citation |
1000+
|
| gptkbp:coauthor |
Heng Wang
Lorenzo Torresani |
| gptkbp:designedFor |
video understanding
|
| gptkbp:developedBy |
gptkb:Facebook_AI_Research
|
| gptkbp:handles |
spatiotemporal data
|
| https://www.w3.org/2000/01/rdf-schema#label |
Timesformer
|
| gptkbp:improves |
previous CNN-based video models
|
| gptkbp:input |
video frames
|
| gptkbp:introducedIn |
2021
|
| gptkbp:language |
gptkb:Python
|
| gptkbp:notablePublication |
Is Space-Time Attention All You Need for Video Understanding?
|
| gptkbp:openSource |
yes
|
| gptkbp:platform |
gptkb:PyTorch
|
| gptkbp:publicationDate |
CVPR 2021
|
| gptkbp:repository |
https://github.com/facebookresearch/TimeSformer
|
| gptkbp:uses |
transformer architecture
divided space-time attention |
| gptkbp:bfsParent |
gptkb:Hugging_Face_models
|
| gptkbp:bfsLayer |
7
|