Activity-driven Weakly-Supervised Spatio-Temporal Grounding from Untrimmed Videos.
Junwen ChenWentao BaoYu KongPublished in: ACM Multimedia (2020)
Keyphrases
- weakly supervised
- spatio temporal
- human activities
- weakly labeled
- object class
- superpixels
- topic models
- video sequences
- relation extraction
- moving objects
- semi supervised
- video frames
- image sequences
- object detectors
- named entities
- video data
- computer vision
- viewpoint
- multiple images
- probabilistic model
- automatic extraction
- object recognition
- natural language
- multiscale
- three dimensional