and achieved 98% accuracy and F1 score on test data. I did the same with a simple CNN and achieved 97% accuracy and F1 score on test data. My question is how is it possible for a simple CNN to achieve results similar to EfficientNet V2 that just has 1 million parameters?
Maybe because of your dataset
My dataset consists of 9000 images. Yeah, I performed preprocessing. How to determine if there is overfitting or not?
Обсуждают сегодня