VisualCheXbert: addressing the discrepancy between radiology report labels and image labels.
Saahil JainAkshay SmitSteven Q. H. TruongChanh D. T. NguyenMinh-Thanh HuynhMudit JainVictoria A. YoungAndrew Y. NgMatthew P. LungrenPranav RajpurkarPublished in: CHIL (2021)
Keyphrases
- image data
- weakly labeled
- input image
- image features
- segmented images
- auto annotation
- confidence scores
- image analysis
- labeling process
- pairwise
- image classification
- image content
- image segmentation
- semantic segmentation
- images with ground truth
- single image
- multi label
- image regions
- image collections
- template matching
- multiscale
- image retrieval
- image representation
- low level
- labeled images
- pixel level
- semantic labels
- ground truth
- edge detection
- feature points
- image matching
- segmentation method
- hough transform
- pixel values
- keypoints
- web images
- image structure
- scale space
- region of interest
- high resolution
- training set