DDLP: Unsupervised Object-Centric Video Prediction with Deep Dynamic Latent Particles.
Tal DanielAviv TamarPublished in: CoRR (2023)
Keyphrases
- prediction accuracy
- object model
- video sequences
- d objects
- event detection
- video data
- real time
- latent variables
- video streams
- dynamic environments
- object detection and tracking
- video frames
- spatial and temporal
- semi supervised
- moving objects
- unsupervised manner
- prediction model
- prediction error
- dynamic textures
- stationary camera
- scalable video coding
- video objects
- deep architectures
- object motion
- multiple objects
- video clips
- partial occlusion
- video surveillance
- object tracking
- image quality
- multimedia