app.py

from flask import Flask
from flask import request
from tensorflow.keras.preprocessing.image import img_to_array
from tensorflow.keras.models import load_model
import numpy as np
import pickle
import cv2
from io import BytesIO
from PIL import Image
import base64
import time

app = Flask(__name__)

# AI 모델 판단 후 결과 응답
@app.route("/model/",methods=['GET', 'POST'])
def decision():
    if(request.method == 'POST'):
        params = request.get_json()['id']
        
        # base64 decode
        img = Image.open(BytesIO(base64.b64decode(params)))
        result = img_to_result(img)
        return result
    
    elif(request.method =='GET'):
        return 'Backend-server Connect'

def img_to_result(image):
    image = np.asarray(image)
    image = cv2.cvtColor(image,cv2.COLOR_RGB2BGR)

    args = {'model': 'p6flower.model', 'labelbin': 'lb6.pickle'}
    
    image = cv2.resize(image, (224,224))
    image = image.astype("float") / 255.0
    
    image = img_to_array(image)
    image = np.expand_dims(image, axis = 0)
    
    model = load_model(args["model"])
    lb = pickle.loads(open(args["labelbin"], "rb").read())
    proba = model.predict(image)[0]
    idx = np.argsort(-proba)[:3]
    label = lb.classes_[idx]

    proba = proba[idx].astype(np.float64)
    proba = np.floor(proba*1000)/10
    
    image_name_list = [{'flower1':label[0],'accuracy1':proba[0]},{'flower2':label[1],'accuracy2':proba[1]},{'flower3':label[2],'accuracy3':proba[2]}]

    return image_name_list

if __name__ == '__main__':
    app.run(host='0.0.0.0' ,port = 5001, debug=True)