Hi, I am a begineer aboul ML
I would like to customize example model from tensorflow website
[图像分类 | TensorFlow Core]
when training finished, I got 99% accuracy but when I using blinding test, it only 65% accuracy, I already use data augmentation and drop out.
Here is my problems
- Should I adding more layer or there is another way to improve accuracy?
- Is there any function or method help you analysis model so I can understand why this picture is false positive or true negative?
Thanks for your information