Pre-Processing for Improving Classification

[Shecyy]

Hello, EnlightenGAN's last experiment validated that it can improve the performance of recognition on the ExDark dataset. ExDark contains 12 object classes (Bicycle, Boat, Bottle, Bus, Car, Cat, Chair, Cup, Dog, Motorbike, Perople, Tabel). I found that the names of these 12 categories do not correspond to the ImageNet categories. For example, there is no “People” category in ImageNet dataset, how do the category names correspond between ExDark and ImageNet?

We very much look forward to hearing from you.