This project consists of the TCCI dataset for the paper "CU-Net: Component Unmixing Network for Textile Fiber Identification".
The TCCI dataset comprises detailed labeling information, including textile categories and composition ratios. In total, 280 textile categories and 68 fiber classes have been cataloged. The distribution of textiles in the dataset is as follows: 2.17% represent single-component textiles, 35.62% as two-component textiles, 53.18% as three-component textiles, 7.36% as four-component textiles, and 1.67% as textiles with other component configurations. Notably, the minimum composition ratio for any component in a four-component textile typically falls below 5%.
The world's first large-scale textile category and component identification dataset for image-based fiber identification, TCCI, can be downloaded at the following link.
- Baiduyun: this link, code: scev