Silicodata: A benchmark chest X-ray dataset for Silicosis detection with radiologist's annotations

This research attempts to provide first its type public dataset for silicosis detection. The dataset contains frontal chest X-rays collected over three years. The data has been collected from the stone workers in the primary health centres of Rajasthan, India. The dataset contains samples of silicosis, STB, TB, and normal. The average size of each sample is around 3567 x 2898. The samples in the dataset are made available as three-channel JPG images. The dataset is divided into two sets:Set A: It contains images with only disease labels.Set B: It contains images with lung segmentation maps, annotations and disease labels