Domain Adaptive Video Semantic Segmentation via Cross-Domain Moving Object Mixing.
Kyusik ChoSuhyeon LeeHongje SeongEuntai KimPublished in: CoRR (2022)
Keyphrases
- cross domain
- semantic segmentation
- video sequences
- moving objects
- transfer learning
- knowledge transfer
- superpixels
- scene classification
- video data
- weakly supervised
- conditional random fields
- target domain
- multiscale
- text categorization
- video frames
- image sequences
- image understanding
- object categories
- object recognition
- object segmentation
- computer vision
- k nearest neighbor
- active learning
- prior knowledge