This dataset has been introduced in the ECCV 2024 paper titled Enriching Information and Preserving Semantic Consistency in Expanding Curvilinear Object Segmentation Datasets.
It encompasses three data types (directories), namely angiography (angiography coronary artery disease), crack, and retina (retinal vessels), which collectively contain six public datasets as described in the paper.
Additionally, the unprocessed_json directory includes raw, unprocessed textual… See the full description on the dataset page:
https://huggingface.co/datasets/456hf/COSTG_v1.
