1 code implementation • 13 Sep 2023 • Fabian Gröger, Simone Lionetti, Philippe Gottfrois, Alvaro Gonzalez-Jimenez, Matthew Groh, Roxana Daneshjou, Labelling Consortium, Alexander A. Navarini, Marc Pouly
Benchmark datasets for digital dermatology unwittingly contain inaccuracies that reduce trust in model performance estimates.
no code implementations • 26 May 2023 • Fabian Gröger, Simone Lionetti, Philippe Gottfrois, Alvaro Gonzalez-Jimenez, Ludovic Amruthalingam, Labelling Consortium, Matthew Groh, Alexander A. Navarini, Marc Pouly
Most benchmark datasets for computer vision contain irrelevant images, near duplicates, and label errors.