Devployment devlop ready to deploy

Dockerfile

@@ -12,6 +12,4 @@ RUN pip install python-multipart
 
 COPY . /code
 
-RUN wget --no-check-certificate 'https://download1980.mediafire.com/ob5057pwitpgMK9TOKsAQBJ0xNRIu5f2vCiNjyPtQjcrM0jcP1gRcGVTCTiDUeQWu1ov55f3eZFPB6bhdsIXoS4WlNNWSGUQRd8jnTrSy82bM9eu16nhpy_rposaaGHxiP62ZbDCOyHvW6xvuPw15BdHbmmh0SUvxciiXrqJ-UmvyhgV/fl67w6t0qx3mk1i/softmax_new_model.h5' -O '/code/models/softmax_new_model.h5'
-
 CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "80"]

controller.py

@@ -1,34 +1,33 @@
 from fastapi import UploadFile
 
 import tensorflow as tf
-from keras.utils import load_img, img_to_array
+from keras.models import load_model
+from keras.utils import img_to_array
 import numpy as np
 from PIL import Image
 
 def preprocessing(fileImage: UploadFile):
-    img = Image.open(fileImage.file).resize((224, 224))
-    img_array = img_to_array(img)
-    img_preprocessed = tf.keras.applications.resnet50.preprocess_input(np.expand_dims(img_array, axis=0))
+    img = Image.open(fileImage.file)
+    img = img.resize((224, 224))
+    img_array = img_to_array(img, dtype=np.float32)
+    img_preprocessed = np.expand_dims(img_array, axis=0)
     return img_preprocessed
 
 def post_preprocessing(img_postpreprocessing):
-    model = tf.keras.models.load_model('./models/softmax_new_model.h5')
-    predictions = model.predict(img_postpreprocessing)
-    predicted_class = np.argmax(predictions, axis=1)[0]
-    class_labels = ['fresh', 'spoiled']
-    predicted_label = class_labels[predicted_class]
-    predicted_probability = predictions[0][predicted_class] * 100
-    return predicted_label, predicted_probability
+    model = load_model('./models/beefy-mobilenetv2.h5')
+    predictions = model.predict(img_postpreprocessing).flatten()
+    predictions = tf.nn.sigmoid(predictions, name='sigmoid')
+    predicted_label = tf.where(predictions < 0.5, 'fresh', 'spoiled').numpy()[0]
+    label_final = bytes.decode(predicted_label, 'utf-8')
+    return label_final
 
-
-def predictMeat(file: UploadFile):
+def inference(file: UploadFile):
     try:
         image = preprocessing(fileImage=file)
-        label, probability = post_preprocessing(img_postpreprocessing=image)
-        response = {
+        label = post_preprocessing(img_postpreprocessing=image)
+        responseBody = {
             'label': label,
-            'pobability': probability
         }
-        return (True, response)
+        return True, responseBody
     except:
-        return (False, None)
+        return False, None

main.py

@@ -1,16 +1,24 @@
 from fastapi import Response, status
 from fastapi import FastAPI, UploadFile
+from fastapi.middleware.cors import CORSMiddleware
 
-from controller import predictMeat
+from controller import inference
 
 app = FastAPI()
 
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=['*'],
+    allow_credentials=False,
+    allow_methods=['GET', 'POST'],
+    allow_headers=["*"],
+)
+
 
 @app.post("/predict/", status_code=200)
 async def predict(fileUpload: UploadFile, response: Response):
-    res_predict = predictMeat(file=fileUpload)
-    if (res_predict[0]):
-        responseBody = res_predict[1]
+    res, responseBody = inference(file=fileUpload)
+    if res:
         return responseBody
     else:
         response.status_code = status.HTTP_500_INTERNAL_SERVER_ERROR