In A single notebook where I tried to explain the best practices to improve Image classification accuracy with proper analysis.
Here I try to explain some useful techniques and cutting edge research ideas in short intutive and experiemntal way some htem are following it's mainly come from Fastai and jeremy Horward.
The follwing things you find inside notebook -
- Transfer learning
- Image resolution tricks
- 1cycle policy
- Test time augmentation
- Class activation maps etc.