Objectron: A Large Scale Dataset of Object-Centric Videos in the Wild with Pose Annotations.
Adel AhmadyanLiangkai ZhangJianing WeiArtsiom AblavatskiMatthias GrundmannPublished in: CoRR (2020)
Keyphrases
- d objects
- trecvid multimedia event detection
- event recognition
- annotation tool
- event detection
- position and orientation
- pose estimation
- video sequences
- weakly labeled
- object geometry
- semantic annotation
- video data
- video frames
- object segmentation
- video surveillance
- crowd sourced
- pose determination
- object representation
- human actions
- million images
- object tracking
- lighting conditions
- pose recovery
- natural language descriptions
- moving objects
- human hand
- cluttered background
- metadata
- video collections
- video dataset
- human pose
- user generated
- target object
- dynamic scenes
- human activities
- partial occlusion