DDLP: Unsupervised Object-centric Video Prediction with Deep Dynamic Latent Particles.
Tal DanielAviv TamarPublished in: Trans. Mach. Learn. Res. (2024)
Keyphrases
- video sequences
- prediction accuracy
- d objects
- video data
- unsupervised manner
- video streams
- supervised learning
- dynamic environments
- video content
- multimedia
- object motion
- complex objects
- neural network
- video analysis
- data objects
- spatial and temporal
- object tracking
- space time
- video surveillance
- video retrieval
- dynamic scenes
- unsupervised learning
- bounding box
- video shots
- dynamic textures
- user centric
- moving objects
- deep learning
- real time
- object detection and tracking