What would be the best type of images that could be used for training which i have seen here and in some other place.I found 2 types of image :
- Cropped images of the classification
- Image with the xml of the cropped content labels
Need help in understanding which would be the accurate and the reason it is used this way mostly with the keras and tensorflow. If there is some other way or method please enlighten me. I’m in need to train dynamic context.
Thanks for your time