MoDA: Leveraging Motion Priors from Videos for Advancing Unsupervised Domain Adaptation in Semantic Segmentation.
Fei PanXu YinSeokju LeeSung-Eui YoonIn So KweonPublished in: CoRR (2023)
Keyphrases
- semantic segmentation
- moving camera
- superpixels
- key frames
- conditional random fields
- motion estimation
- label transfer
- image sequences
- video sequences
- object categories
- weakly supervised
- optical flow
- dynamic textures
- dynamic scenes
- moving objects
- human motion
- video data
- scene classification
- image understanding
- video frames
- pixel level
- computer vision
- object recognition
- object segmentation
- action recognition
- input image
- pascal voc
- high resolution
- image segmentation