Update app.py
Browse files
app.py
CHANGED
@@ -1,97 +1,138 @@
|
|
1 |
-
import
|
|
|
2 |
import cv2
|
3 |
import numpy as np
|
4 |
import time
|
5 |
-
# Larger title
|
6 |
-
st.markdown("<h1 style='text-align: center;'>Emotion Detection</h1>", unsafe_allow_html=True)
|
7 |
-
|
8 |
-
# Smaller subtitle
|
9 |
-
st.markdown("<h3 style='text-align: center;'>angry, fear, happy, neutral, sad, surprise</h3>", unsafe_allow_html=True)
|
10 |
-
start = time.time()
|
11 |
from keras.models import load_model
|
12 |
-
import tempfile
|
13 |
from PIL import Image
|
|
|
14 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
@st.cache_resource
|
16 |
def load_emotion_model():
|
17 |
-
|
18 |
-
return model
|
19 |
|
20 |
-
|
21 |
-
model = load_emotion_model()
|
22 |
-
print("time taken to load model : " , time.time() - start)
|
23 |
-
img_shape = 48
|
24 |
-
emotion_labels = ['angry', 'fear', 'happy', 'neutral', 'sad', 'surprise']
|
25 |
-
face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
|
26 |
|
|
|
|
|
|
|
|
|
27 |
|
28 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
gray_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
|
|
|
30 |
faces = face_cascade.detectMultiScale(gray_frame, scaleFactor=1.1, minNeighbors=5, minSize=(30, 30))
|
31 |
|
32 |
for (x, y, w, h) in faces:
|
33 |
-
|
34 |
-
|
|
|
35 |
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
emotion = emotion_labels[np.argmax(predictions[0])]
|
41 |
|
|
|
|
|
|
|
|
|
|
|
42 |
cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 2)
|
43 |
-
cv2.putText(frame,
|
44 |
|
45 |
return frame
|
46 |
|
47 |
-
# def video_feed(video_source):
|
48 |
-
# # Read and process video frames
|
49 |
-
# while True:
|
50 |
-
# ret, frame = video_source.read()
|
51 |
-
# if not ret:
|
52 |
-
# break
|
53 |
-
# frame = process_frame(frame)
|
54 |
-
# st.image(frame, channels="BGR")
|
55 |
-
|
56 |
def video_feed(video_source):
|
57 |
-
|
58 |
-
|
|
|
|
|
59 |
|
60 |
while True:
|
61 |
ret, frame = video_source.read()
|
62 |
if not ret:
|
63 |
break
|
64 |
|
65 |
-
frame =
|
66 |
-
|
67 |
-
# Display the frame in the placeholder
|
68 |
frame_placeholder.image(frame, channels="BGR", use_column_width=True)
|
69 |
|
70 |
-
|
71 |
-
|
72 |
-
# Sidebar for video or image upload
|
73 |
-
upload_choice = st.sidebar.radio("Choose input source", [ "Upload Video", "Upload Image" ,"Camera"])
|
74 |
|
75 |
if upload_choice == "Camera":
|
76 |
-
# Access
|
77 |
-
video_source = cv2.VideoCapture(0)
|
78 |
video_feed(video_source)
|
79 |
|
80 |
elif upload_choice == "Upload Video":
|
81 |
uploaded_video = st.file_uploader("Upload Video", type=["mp4", "mov", "avi", "mkv", "webm"])
|
82 |
if uploaded_video:
|
83 |
-
# Temporarily save the video to disk
|
84 |
with tempfile.NamedTemporaryFile(delete=False) as tfile:
|
85 |
tfile.write(uploaded_video.read())
|
86 |
video_source = cv2.VideoCapture(tfile.name)
|
87 |
video_feed(video_source)
|
88 |
|
89 |
elif upload_choice == "Upload Image":
|
90 |
-
uploaded_image = st.file_uploader("Upload Image", type=["png", "jpg", "jpeg"
|
91 |
if uploaded_image:
|
92 |
image = Image.open(uploaded_image)
|
93 |
frame = np.array(image)
|
94 |
-
frame =
|
95 |
-
st.image(frame, caption=
|
96 |
-
|
97 |
-
st.sidebar.write("Emotion Labels: Angry, Fear, Happy, Neutral, Sad, Surprise")
|
|
|
1 |
+
import face_recognition
|
2 |
+
import os
|
3 |
import cv2
|
4 |
import numpy as np
|
5 |
import time
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
from keras.models import load_model
|
|
|
7 |
from PIL import Image
|
8 |
+
import streamlit as st
|
9 |
|
10 |
+
# Streamlit UI Setup
|
11 |
+
st.markdown("<h1 style='text-align: center;'>Emotion & Face Recognition</h1>", unsafe_allow_html=True)
|
12 |
+
st.markdown("<h3 style='text-align: center;'>angry, fear, happy, neutral, sad, surprise</h3>", unsafe_allow_html=True)
|
13 |
+
|
14 |
+
# Known faces folder path
|
15 |
+
KNOWN_FACES_DIR = "known_faces"
|
16 |
+
|
17 |
+
# Load emotion detection model
|
18 |
@st.cache_resource
|
19 |
def load_emotion_model():
|
20 |
+
return load_model("CNN_Model_acc_75.h5")
|
|
|
21 |
|
22 |
+
emotion_model = load_emotion_model()
|
|
|
|
|
|
|
|
|
|
|
23 |
|
24 |
+
# Face detection model
|
25 |
+
face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + "haarcascade_frontalface_default.xml")
|
26 |
+
emotion_labels = ['angry', 'fear', 'happy', 'neutral', 'sad', 'surprise']
|
27 |
+
img_shape = 48
|
28 |
|
29 |
+
# Known faces dictionary
|
30 |
+
known_faces = {"names": [], "encodings": []}
|
31 |
+
|
32 |
+
def load_faces_from_folder(folder_path):
|
33 |
+
"""
|
34 |
+
Load known faces from a folder, using filenames as names.
|
35 |
+
"""
|
36 |
+
for filename in os.listdir(folder_path):
|
37 |
+
if filename.endswith(('.jpg', '.jpeg', '.png')):
|
38 |
+
name = os.path.splitext(filename)[0]
|
39 |
+
image_path = os.path.join(folder_path, filename)
|
40 |
+
|
41 |
+
# Load and encode the image
|
42 |
+
image = face_recognition.load_image_file(image_path)
|
43 |
+
face_encodings = face_recognition.face_encodings(image)
|
44 |
+
|
45 |
+
if face_encodings: # Ensure a face is found
|
46 |
+
known_faces["names"].append(name)
|
47 |
+
known_faces["encodings"].append(face_encodings[0])
|
48 |
+
print(f"Loaded face for {name}")
|
49 |
+
else:
|
50 |
+
print(f"No face detected in {filename}")
|
51 |
+
|
52 |
+
# Load known faces
|
53 |
+
load_faces_from_folder(KNOWN_FACES_DIR)
|
54 |
+
|
55 |
+
def recognize_face(unknown_face_encoding):
|
56 |
+
"""
|
57 |
+
Compare an unknown face with the known faces and return the closest match.
|
58 |
+
"""
|
59 |
+
matches = face_recognition.compare_faces(known_faces["encodings"], unknown_face_encoding, tolerance=0.6)
|
60 |
+
if True in matches:
|
61 |
+
match_index = matches.index(True)
|
62 |
+
return known_faces["names"][match_index]
|
63 |
+
return "Unknown"
|
64 |
+
|
65 |
+
def detect_emotion(face_image):
|
66 |
+
"""
|
67 |
+
Predict the emotion of a face using the emotion detection model.
|
68 |
+
"""
|
69 |
+
face_resized = cv2.resize(face_image, (img_shape, img_shape))
|
70 |
+
face_resized = np.expand_dims(face_resized, axis=0)
|
71 |
+
face_resized = face_resized / 255.0 # Normalize the image
|
72 |
+
predictions = emotion_model.predict(face_resized)
|
73 |
+
return emotion_labels[np.argmax(predictions)]
|
74 |
+
|
75 |
+
def process_frame_with_recognition_and_emotion(frame):
|
76 |
+
"""
|
77 |
+
Detect faces, recognize names, and detect emotions in the frame.
|
78 |
+
"""
|
79 |
gray_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
|
80 |
+
rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
|
81 |
faces = face_cascade.detectMultiScale(gray_frame, scaleFactor=1.1, minNeighbors=5, minSize=(30, 30))
|
82 |
|
83 |
for (x, y, w, h) in faces:
|
84 |
+
# Get the face area
|
85 |
+
face_image = rgb_frame[y:y+h, x:x+w]
|
86 |
+
face_encodings = face_recognition.face_encodings(face_image)
|
87 |
|
88 |
+
if face_encodings:
|
89 |
+
name = recognize_face(face_encodings[0]) # Recognize the face
|
90 |
+
else:
|
91 |
+
name = "Unknown"
|
|
|
92 |
|
93 |
+
# Predict emotion
|
94 |
+
emotion = detect_emotion(frame[y:y+h, x:x+w])
|
95 |
+
|
96 |
+
# Display name and emotion
|
97 |
+
display_text = f"{name} is Feeling {emotion}"
|
98 |
cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 2)
|
99 |
+
cv2.putText(frame, display_text, (x, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (0, 255, 0), 2)
|
100 |
|
101 |
return frame
|
102 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
103 |
def video_feed(video_source):
|
104 |
+
"""
|
105 |
+
Display video feed with face recognition and emotion detection.
|
106 |
+
"""
|
107 |
+
frame_placeholder = st.empty() # Placeholder for updating frames
|
108 |
|
109 |
while True:
|
110 |
ret, frame = video_source.read()
|
111 |
if not ret:
|
112 |
break
|
113 |
|
114 |
+
frame = process_frame_with_recognition_and_emotion(frame)
|
|
|
|
|
115 |
frame_placeholder.image(frame, channels="BGR", use_column_width=True)
|
116 |
|
117 |
+
# Sidebar options
|
118 |
+
upload_choice = st.sidebar.radio("Choose input source", ["Upload Image", "Upload Video", "Camera"])
|
|
|
|
|
119 |
|
120 |
if upload_choice == "Camera":
|
121 |
+
video_source = cv2.VideoCapture(0) # Access webcam
|
|
|
122 |
video_feed(video_source)
|
123 |
|
124 |
elif upload_choice == "Upload Video":
|
125 |
uploaded_video = st.file_uploader("Upload Video", type=["mp4", "mov", "avi", "mkv", "webm"])
|
126 |
if uploaded_video:
|
|
|
127 |
with tempfile.NamedTemporaryFile(delete=False) as tfile:
|
128 |
tfile.write(uploaded_video.read())
|
129 |
video_source = cv2.VideoCapture(tfile.name)
|
130 |
video_feed(video_source)
|
131 |
|
132 |
elif upload_choice == "Upload Image":
|
133 |
+
uploaded_image = st.file_uploader("Upload Image", type=["png", "jpg", "jpeg"])
|
134 |
if uploaded_image:
|
135 |
image = Image.open(uploaded_image)
|
136 |
frame = np.array(image)
|
137 |
+
frame = process_frame_with_recognition_and_emotion(frame)
|
138 |
+
st.image(frame, caption="Processed Image", use_column_width=True)
|
|
|
|