models
models copied to clipboard
speedup image classification processing
Speedup Image Classification training speed
- [x] add eval shape arguments, for different image shape processing, like inception v4
- [x] add multiprocess reader, set cv2 as the default reader
- [x] call reading files func in every training loop, and use pass id as the seed to implement global shuffle
- [x] run normalization operations onto GPU, use uint8 to transfer data in order to use fewer bytes
Experiments Results: