Update app.py

app.py

@@ -1,10 +1,12 @@
 import base64
 import cv2
 import numpy as np
+import pytesseract
 from flask import Flask, request, jsonify, render_template 
 from tensorflow import keras
 import sys
 import io
+from inference_sdk import InferenceHTTPClient
 
 # Set the default encoding to utf-8
 sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')
@@ -12,7 +14,14 @@ sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')
 app = Flask(__name__)
 
 # Load your pre-trained model
-model = keras.models.load_model('fresh_model.keras')
+fresh_model = keras.models.load_model('fresh_model.keras')
+
+# Initialize the InferenceHTTPClient for the external model API
+CLIENT = InferenceHTTPClient(
+    api_url="https://detect.roboflow.com",
+    api_key="QDeIvOVhC1LVQhp1wRti"
+)
+
 vegetables = [
     "banana", "beans broad", "beans cluster", "beans haricot", "beetroot",
     "bitter guard", "bottle guard", "brinjal long", "brinjal[purple]", "cabbage",
@@ -39,25 +48,46 @@ def predict():
         image = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
 
         # Convert to RGB format
-        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+        rgb_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+
+        # Detect text using Tesseract OCR
+        text = pytesseract.image_to_string(rgb_image)
+
+        # Check if any text is detected
+        if text.strip():
+            print(f"Detected text: {text}")
+            # Use the fresh_model for prediction if text is detected
+
+            # Resize the image to the expected input size of the model (e.g., 225x225)
+            image_resized = cv2.resize(rgb_image, (225, 225))
+
+            # Normalize the image
+            image_resized = image_resized.astype('float32') / 255.0  # Normalize to [0, 1]
+            image_resized = np.expand_dims(image_resized, axis=0)  # Add batch dimension
+
+            # Make a prediction using the model
+            predictions = fresh_model.predict(image_resized)
+            predicted_class = np.argmax(predictions, axis=1)[0]
 
-        # Resize the image to the expected input size of the model (e.g., 225x225)
-        image = cv2.resize(image, (225, 225))
+            # Map the class index to the vegetable label
+            prediction_label = vegetables[predicted_class]
 
-        # Normalize the image
-        image = image.astype('float32') / 255.0  # Normalize to [0, 1]
-        image = np.expand_dims(image, axis=0)  # Add batch dimension
+            return jsonify({'prediction': prediction_label})
+        else:
+            print("No text detected. Using external model inference.")
+            # Convert the image back to bytes for API use
+            _, buffer = cv2.imencode('.jpg', rgb_image)
+            image_bytes = buffer.tobytes()
 
-        # Make a prediction using the model
-        predictions = model.predict(image)
-        predicted_class = np.argmax(predictions, axis=1)[0]
+            # Use the InferenceHTTPClient to send the image to the external API
+            result = CLIENT.infer(image_bytes, model_id="zydus-wellness-acjbd/2")
 
-        # Map the class index to the vegetable label
-        prediction_label = vegetables[predicted_class]
+            # Process the result and extract necessary information
+            prediction_label = result.get('predictions', [{}])[0].get('label', 'Unknown')
+            return jsonify({'prediction': prediction_label})
 
-        return jsonify({'prediction': prediction_label})
     except Exception as e:
         return jsonify({'error': str(e)}), 500
 
 if __name__ == '__main__':
-    app.run(host='0.0.0.0', port=7860, debug=True)
+    app.run(host='0.0.0.0', port=7860, debug=True)