Detection of Distress Region from Subway Tunnel Images via U-net-based Deep Semantic Segmentation.
An WangRen TogoTakahiro OgawaMiki HaseyamaPublished in: GCCE (2019)
Keyphrases
- semantic segmentation
- street scenes
- input image
- image data
- region of interest
- label transfer
- superpixels
- ground truth
- image classification
- image features
- image database
- image regions
- test images
- image retrieval
- three dimensional
- bounding box
- image understanding
- image set
- multiple images
- object recognition
- pascal voc
- lighting conditions
- natural images
- image content
- partial occlusion
- hidden markov models
- scene classification
- weakly supervised
- training set
- segmentation method