Quantifying the Effects of Ground Truth Annotation Quality on Object Detection and Instance Segmentation Performance.
Cathaoir AgnewCiarán EisingPatrick DennyAnthony G. ScanlanPepijn van de VenEoin Martino GruaPublished in: IEEE Access (2023)
Keyphrases
- ground truth
- object detection
- ground truth data
- high quality
- object segmentation
- image segmentation algorithms
- segmented images
- image parsing
- image segmentation
- level set
- segmentation algorithm
- segmentation method
- quantitative evaluation
- medical images
- shape prior
- scene understanding
- semantic annotation
- metadata
- face detection
- test images
- region growing
- contour detection
- medical imaging
- quality assessment
- background subtraction
- computer vision
- image regions
- segmentation accuracy
- automatic image annotation
- image classification
- scene recognition
- image segmentation algorithm
- object categories
- image analysis
- manually labeled