ICASSP 2024 — TransMix
Cross-domain augmentation for limited-supervision foot-ulcer segmentation. ICASSP 2024 (equal first author).
Improving Limited Supervised Foot Ulcer Segmentation Using Cross-Domain Augmentation Strategies. Equal first author with Po-Han Huang. Published at IEEE ICASSP 2024.
A two-stage cross-domain augmentation method (TransMix) combining Augmented Global Pre-training on HAM10000 with Localized CutMix Fine-tuning on FUSeg. Carried out at Inventec AI Center in collaboration with Taipei Veterans General Hospital.
See the Publications page for the full bibliographic entry, abstract, and downloads (IEEE Xplore, arXiv, BibTeX). The publication card is the authoritative source for the paper’s results and links — this project page exists as a brief index entry under Projects.