I am currently using working on training image model for classifying real, synthetic, semi-synthetic images. I am using transfer learning on dinov3-vit-large model. I have data corpus about 330k images. In the phase 1 experiment, I trained the model to classify real vs synthetic(synthetic + semi-synthetic). Now in phase 2, I want the model to classify real vs synthetic vs semi-synthetic. But the biggest concern is the dataset balance. Currently...
Originally asked by fantasywood on Stack Exchange (ai) (CC BY-SA 4.0).
Variant 1055.