Update model and draw boundary

app.py

@@ -1,45 +1,117 @@
 import cv2
 import streamlit as st
-from keras_vggface.vggface import VGGFace
-import tensorflow as tf
-from tensorflow.keras.applications import ResNet50
-from tensorflow.keras.preprocessing import image
-from tensorflow.keras.models import Model
-from tensorflow.keras.layers import GlobalAveragePooling2D
+from PIL import Image
 import numpy as np
-import pickle
-
-pickle_file_path = 'models/svm_model.pkl'
+from tensorflow.keras.preprocessing import image
+from tensorflow.keras.applications.vgg16 import preprocess_input
+import tensorflow as tf
+from base64 import b64decode, b64encode
+import PIL
+import io
+import html
+import time
 
-with open(pickle_file_path, 'rb') as file:
-    svm_model = pickle.load(file)
 
-base_model = VGGFace(model='vgg16', include_top=False, input_shape=(224, 224, 3))
-x = base_model.output
-x = GlobalAveragePooling2D()(x)
-model = Model(inputs=base_model.input, outputs=x)
+model_file_path = 'models/bmi.h5'
+model = tf.keras.models.load_model(model_file_path)
 
-# Function to preprocess the image
-def preprocess_image(img):
-    img = cv2.resize(img, (224, 224))
+# Preprocess the images for VGG16
+def preprocess_image(img_path):
+    img = image.load_img(img_path, target_size = (224, 224))
     img = image.img_to_array(img)
-    img = np.expand_dims(img, axis=0)
-    img = img[0] # Remove the extra dimension
+    img = np.expand_dims(img, axis = 0)
+    img = preprocess_input(img)
     return img
+
+
+# function to convert OpenCV Rectangle bounding box image into base64 byte string to be overlayed on video stream
+def bbox_to_bytes(bbox_array):
+  """
+  Params:
+          bbox_array: Numpy array (pixels) containing rectangle to overlay on video stream.
+  Returns:
+        bytes: Base64 image byte string
+  """
+  # convert array into PIL image
+  bbox_PIL = PIL.Image.fromarray(bbox_array, 'RGBA')
+  iobuf = io.BytesIO()
+  # format bbox into png for return
+  bbox_PIL.save(iobuf, format='png')
+  # format return string
+  bbox_bytes = 'data:image/png;base64,{}'.format((str(b64encode(iobuf.getvalue()), 'utf-8')))
+
+  return bbox_bytes
+
+# base_model = VGGFace(model='vgg16', include_top=False, input_shape=(224, 224, 3))
+# x = base_model.output
+# x = GlobalAveragePooling2D()(x)
+# model = Model(inputs=base_model.input, outputs=x)
+
+# # Function to preprocess the image
+# def preprocess_image(img):
+#     img = cv2.resize(img, (224, 224))
+#     img = image.img_to_array(img)
+#     img = np.expand_dims(img, axis=0)
+#     img = img[0] # Remove the extra dimension
+#     return img
     
-def extract_features(image_array):
-    # img = np.squeeze(image_array, axis=0)
-    img = np.expand_dims(image_array, axis=0)
-    img = tf.keras.applications.resnet50.preprocess_input(img)
-    features = model.predict(img,verbose=0)
-    return features.flatten()
+# def extract_features(image_array):
+#     # img = np.squeeze(image_array, axis=0)
+#     img = np.expand_dims(image_array, axis=0)
+#     img = tf.keras.applications.resnet50.preprocess_input(img)
+#     features = model.predict(img,verbose=0)
+#     return features.flatten()
 
 # Function to predict BMI
+
+def draw_boundary(img):
+    # initialize the Haar Cascade face detection model
+    face_cascade = cv2.CascadeClassifier(cv2.samples.findFile(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml'))
+
+    # initialze bounding box to empty
+    bbox = ''
+    count = 0 
+    while True:
+
+        # create transparent overlay for bounding box
+        bbox_array = np.zeros([480,640,4], dtype=np.uint8)
+
+        # grayscale image for face detection
+        gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
+
+        # get face region coordinates
+        faces = face_cascade.detectMultiScale(gray)
+        # get face bounding box for overlay
+        for (x, y, w, h) in faces:
+            # Extract the face region from the frame
+            face = img[y:y+h, x:x+w]
+            
+            # Preprocess the face image
+            face = cv2.resize(face, (224, 224))
+            face = cv2.cvtColor(face, cv2.COLOR_BGR2RGB)
+            face = preprocess_input(face)/255.
+            face = np.expand_dims(face, axis=0)
+            
+            # Predict BMI using the pre-trained model
+            bmi = model.predict(face)[0][0]
+            
+            # Draw the predicted BMI on the frame
+            bbox_array = cv2.putText(bbox_array, f'BMI: {bmi:.2f}', (x, y-10), cv2.FONT_HERSHEY_SIMPLEX, 0.9, (0, 255, 0), 2)
+            
+            # Draw a rectangle around the face
+            bbox_array = cv2.rectangle(bbox_array, (x, y), (x+w, y+h), (255, 0, 0), 2)
+
+        bbox_array[:,:,3] = (bbox_array.max(axis = 2) > 0 ).astype(int) * 255
+        # convert overlay of bbox into bytes
+        bbox_bytes = bbox_to_bytes(bbox_array)
+        # update bbox so next frame gets new overlay
+        bbox = bbox_bytes
+    
+    return img
+
 def predict_bmi(img):
     pre_img = preprocess_image(img)
-    features = extract_features(pre_img)
-    features = features.reshape(1,-1)
-    pred = svm_model.predict(features)
+    pred = draw_boundary(pre_img)
     return pred
 
 def main():

models/bmi.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72043cba1fe1ab39ff72bf8581ed4547a9506376f2befdbd680745984684403f
+size 136023928