# -*- coding: utf-8 -*-
"""app.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1F6tnDcfVlChPXwu4UwSHq-bggJlcmyj-

# This notebook makes predictions based on saved model.

#Import Libraries
"""
#|default_exp app

import sys
import subprocess

# implement pip as a subprocess:
subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'fastai'])
subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'fastbook'])
subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'gradio'])
subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'nbdev'])
# process output with an API in the subprocess module:
reqs = subprocess.check_output([sys.executable, '-m', 'pip', 'freeze'])
installed_packages = [r.decode().split('==')[0] for r in reqs.split()]
print(installed_packages)

'''
#|export
!pip install -Uqq fastai
!pip install -Uqq fastbook
!pip install -Uqq gradio
!pip install -Uqq nbdev
'''
"""#Dogs Vs Cats"""

#|export
from fastai.vision.all import *
import gradio as gr

#|export
def is_cat(x): return x[0].isupper()

'''
im = PILImage.create('/content/dog.JPG')
im.thumbnail((192,192))
im
'''
#|export
#Use the model.pkl file that is the classifier model
learn = load_learner('model.pkl')

# Commented out IPython magic to ensure Python compatibility.
#Now predicted the image to determine if it is a Cat ? It correctly predicts that it is NOT a Cat.
#learn.predict(im)

# %time learn.predict(im)

"""#Now create a GRADIO Interface that has this information

Gradio requires us to give it a function that it is going to call. Our's function is `classify_image(img)`

learn.predict(img) returns 3 things :
pred --> Prediction as a String
idx --> Its index
probs --> Probability that the image is a Cat ?

Gradio wants to get back a dictionary containing each of the posible categories -- which is this case is a DOg or a Cat -- and the probability of each one..

"""

#|export
categories = ('Dog','Cat')

def classify_image(img):
  pred,idx,probs = learn.predict(img)
  return dict(zip(categories,map(float,probs)))

#classify_image(im)

"""#Now creating our GRADIO Interface"""

#|export
image = gr.Image(height=192,width=192)
label = "label"
examples = ['dog.JPG','cat.JPG','dunno.JPG']
intf = gr.Interface(fn = classify_image,inputs=image,outputs=label,examples=examples)
intf.launch(inline=False)

"""#Export the notebook as Python script"""
'''
!pip install -Uqq nbdev

#from nbdev.export import notebook2script

from nbdev import nbdev_export

!pip install -Uqq nbconvert

!jupyter nbconvert app.ipynb --to python
'''