try diff image preprocessing before predictions

app.py

@@ -6,25 +6,40 @@ import numpy as np
 
 
 # Load the saved model
-model = tf.keras.models.load_model('model/cnn_9_layer_model.h5')
+model = tf.keras.models.load_model('model/model.h5')
 
 # Define the face cascade and emotions
 face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
 emotions = ['Angry', 'Disgust', 'Fear', 'Happy', 'Sad', 'Surprise', 'Neutral']
 
+no_face_detection_alert = "Cannot Detect Face"
+low_confidence_alert = "Cannot Detect Emotion"
+
 # Define the predict_emotion function
-def predict_emotion(frame):
-	gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
-	faces = face_cascade.detectMultiScale(gray, 1.3, 5)
-	for (x, y, w, h) in faces:
-		face = gray[y:y+h, x:x+w]
-		face = cv2.resize(face, (48, 48))
-		face = np.expand_dims(face, axis=-1)
-		face = np.expand_dims(face, axis=0)
-		prediction = model.predict(face)
-		emotion = emotions[np.argmax(prediction)]
-		cv2.putText(frame, emotion, (x, y-10), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)
-		cv2.rectangle(frame, (x, y), (x+w, y+h), (255, 0, 0), 2)
+    for (x, y, w, h) in faces:
+            face = gray[y:y+h, x:x+w]
+            face = cv2.resize(face, (48, 48), interpolation = cv2.INTER_AREA)
+            if np.sum([face])!=0:
+                face_test = face.astype('float')/255.0
+                face_test = tf.keras.utils.img_to_array(face_test)
+                face_test = np.expand_dims(face_test, axis=0)
+                prediction = model.predict(face_test)
+
+
+                if any(prob >.5 for prob in prediction[0]):
+
+                    emotion = emotions[np.argmax(prediction)]
+
+                    cv2.putText(frame, emotion, (x, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 1, (128, 128, 0), 2)
+                    cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 0, 0), 2)
+
+                else:
+                    cv2.putText(frame, low_confidence_alert, (x, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 69, 255),
+                                2)
+
+            else:
+                cv2.putText(frame, no_face_detection_alert, (x, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 69, 255), 2)
+
 
 	return frame
 
@@ -44,7 +59,7 @@ def predict_emotion(frame):
 input_image = gr.Image(source = "webcam", streaming = True, label="Your Face")
 # video = gr.inputs.Video(source = "webcam" )
 
-output_image = gr.Image( type="pil", label="Detected Emotion" )
+output_image = gr.Image( type="numpy", label="Detected Emotion" )