Abstract:
Deep learning in histopathology requires large training datasets. Pathologists spend a lot of time labeling virtual slides, even though whole slide imaging scanners speed up data acquisition. Eye gaze annotations may speed slide labeling. This study compares the efficacy and timing of eye gaze labeling to manual labeling for training object detectors. Discussed are gaze-based labeling challenges and methods to refine coarse data annotations for object detection. Gaze tracking-based labeling saves pathologists time and trains a deep object detector well. We compare deep object detectors trained on hand-labelled and gaze-labelled data by localizing Keratin Pearls in oral squamous cell carcinoma patients. Gaze-labeling took 57.6% less time per label than “Bounding-box” labeling and 85% less than “Freehand” labeling.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
