MemSeg: A semi-supervised method for image surface defect detection using differences and commonalities.
Minghui YangPeng WuJing LiuHui FengPublished in: CoRR (2022)
Keyphrases
- segmentation method
- reconstruction method
- semi supervised
- template matching
- surface model
- similarity measure
- pairwise
- high resolution
- test images
- image data
- segmentation algorithm
- edge detection
- image features
- active learning
- surface shape
- energy function
- textured surfaces
- machine learning
- single image
- image classification
- multiscale
- three dimensional
- image regions
- image content
- vector field
- region of interest
- pixel values
- input image
- defect detection
- gradient field
- volumetric data