Depth-Adapted CNNs for RGB-D Semantic Segmentation.
Zongwei WuGuillaume AllibertChristophe StolzChao MaCédric DemonceauxPublished in: CoRR (2022)
Keyphrases
- semantic segmentation
- street scenes
- conditional random fields
- depth information
- label transfer
- superpixels
- scene classification
- weakly supervised
- depth estimation
- depth map
- object classes
- object segmentation
- pascal voc
- object categories
- object class
- image understanding
- long range
- stereo vision
- object detection
- input image
- training data
- biologically inspired
- pixel level
- single image
- natural images
- viewpoint
- object recognition