GillHistDB: A Histological Image Dataset for Fish Gill Lesion Segmentation

Description:This dataset contains RGB histological images of fish gills annotated for semantic segmentation of epithelial lifting (EL) and hyperplasia (HY) lesions. Classes:0 - Background1 - Epithelial lifting (EL)2 - Hyperplasia (HY) Dataset format:The annotations are provided in COCO Segmentation format. Pixel-level masks may also be provided for semantic segmentation experiments. Number of images:447 RGB histological images. Number of annotated lesions:29,730 lesions, including 16,855 epithelial lifting instances and 12,875 hyperplasia instances. Recommended citation: Waiting approve paper.