Digital-Dermatology/CleanPatrick
CleanPatrick: A Benchmark for Data Cleaning Welcome to CleanPatrick, the first large-scale benchmark designed for data cleaning in the image domain. Built on the Fitzpatrick17k dermatology dataset, CleanPatrick is a dataset for measuring the performance in detecting three major data quality issues: off-topic samples, near-duplicates, and label errors. Overview CleanPatrick consists of dermatological images annotated with over 500,000 binary labels across three… See the full description on the dataset page: https://huggingface.co/datasets/Digital-Dermatology/CleanPatrick.
Upload near_duplicates_meta.csv
Upload README.md
Upload README.md
Delete assets/benchmark_results_barplot_top2_with_ci.pdf
Upload benchmark_results_barplot_top2_with_ci.png
Update README.md
Update README.md
Delete assets/benchmark_results_barplot.png
Upload benchmark_results_barplot_top2_with_ci.pdf
Upload 3 files
Upload 7 files
Upload CleanPatrick_Teaser_compressed.png
Update README.md
Upload 7 files
Update README.md
Update README.md
Update README.md
Update README.md
Update README.md
Update README.md
removed PDF assets
Upload CleanPatrick_Teaser_compressed.png
Upload CleanPatrick_Teaser_compressed.pdf
Update README.md
removed PDF assets
Update README.md
Upload 6 files
Upload 3 files
Initial commit