Hacked the resnet to predict 120 classes
usage :
- py dogs.py pilou.jpg # => breed + probability
- py dogs.py &;#runs flask on 127.0.0.1:8080 in background, model gets preloaded => better response times
then perform your postdata using curl such as :
- curl -k 127.0.0.1:8080 -F "input=unexpected";#returns ko
- curl -k 127.0.0.1:8080 -F "filepath=pilou.jpg";#samoyed
- curl -k 127.0.0.1:8080 -F "[email protected]";#samoyed