Upload 3 files
Browse files- app.py +110 -0
- model100e.pt +3 -0
- requirements.txt +8 -0
app.py
ADDED
@@ -0,0 +1,110 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import cv2
|
2 |
+
import torch
|
3 |
+
import torch.backends.cudnn as cudnn
|
4 |
+
from models.experimental import attempt_load
|
5 |
+
from utils.general import non_max_suppression
|
6 |
+
from torchvision import models
|
7 |
+
from torchvision import transforms
|
8 |
+
from PIL import Image
|
9 |
+
import time
|
10 |
+
import streamlit as st
|
11 |
+
|
12 |
+
yolov5_weight_file = 'model100e.pt'
|
13 |
+
|
14 |
+
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
15 |
+
yolov5_model = attempt_load(yolov5_weight_file, device=device, inplace=True, fuse=True)
|
16 |
+
cudnn.benchmark = True
|
17 |
+
names = yolov5_model.module.names if hasattr(yolov5_model, 'module') else yolov5_model.names
|
18 |
+
|
19 |
+
conf_set = 0.1
|
20 |
+
frame_size = (800, 480)
|
21 |
+
|
22 |
+
colors = {
|
23 |
+
'helmet': (255, 0, 0),
|
24 |
+
'rider': (0, 255, 0),
|
25 |
+
'number': (0, 0, 255),
|
26 |
+
'no_helmet': (0, 100, 255),
|
27 |
+
}
|
28 |
+
|
29 |
+
def detect_objects(frame):
|
30 |
+
img = torch.from_numpy(frame)
|
31 |
+
img = img.permute(2, 0, 1).float().to(device)
|
32 |
+
img /= 255.0
|
33 |
+
if img.ndimension() == 3:
|
34 |
+
img = img.unsqueeze(0)
|
35 |
+
with torch.no_grad():
|
36 |
+
pred = yolov5_model(img, augment=False)[0]
|
37 |
+
pred = non_max_suppression(pred, conf_set, 0.30)
|
38 |
+
detections = []
|
39 |
+
for det in pred:
|
40 |
+
if len(det):
|
41 |
+
for d in det: # d = (x1, y1, x2, y2, conf, cls)
|
42 |
+
x1 = int(d[0].item())
|
43 |
+
y1 = int(d[1].item())
|
44 |
+
x2 = int(d[2].item())
|
45 |
+
y2 = int(d[3].item())
|
46 |
+
conf = round(d[4].item(), 2)
|
47 |
+
c = int(d[5].item())
|
48 |
+
detected_name = names[c]
|
49 |
+
detections.append((x1, y1, x2, y2, conf, detected_name))
|
50 |
+
|
51 |
+
color = colors.get(detected_name, (255, 255, 255))
|
52 |
+
cv2.rectangle(frame, (x1, y1), (x2, y2), color, 2)
|
53 |
+
cv2.putText(frame, detected_name, (x1, y1), cv2.FONT_HERSHEY_DUPLEX, 1, color, 2)
|
54 |
+
|
55 |
+
return detections
|
56 |
+
|
57 |
+
def app():
|
58 |
+
st.title("Helmet Detection App")
|
59 |
+
st.write("This app uses YOLOv5 to detect helmets and riders in images and videos.")
|
60 |
+
|
61 |
+
# Select input type
|
62 |
+
input_type = st.radio("Select input type:", options=["Image", "Video"])
|
63 |
+
|
64 |
+
# Upload file or use webcam
|
65 |
+
if input_type == "Image":
|
66 |
+
uploaded_file = st.file_uploader("Upload image", type=["jpg", "jpeg", "png"])
|
67 |
+
if uploaded_file is not None:
|
68 |
+
image = Image.open(uploaded_file)
|
69 |
+
st.image(image, caption="Uploaded Image", use_column_width=True)
|
70 |
+
detections = detect_objects(np.array(image))
|
71 |
+
display_detections(image, detections)
|
72 |
+
|
73 |
+
elif input_type == "Video":
|
74 |
+
st.write("Select an option to get the input video:")
|
75 |
+
video_option = st.radio("", options=["Webcam", "Upload video"])
|
76 |
+
|
77 |
+
if video_option == "Webcam":
|
78 |
+
cap = cv2.VideoCapture(0)
|
79 |
+
elif video_option == "Upload video":
|
80 |
+
uploaded_file = st.file_uploader("Upload video", type=["mp4"])
|
81 |
+
if uploaded_file is not None:
|
82 |
+
temp_file = NamedTemporaryFile(delete=False)
|
83 |
+
temp_file.write(uploaded_file.read())
|
84 |
+
st.write("Video uploaded successfully!")
|
85 |
+
cap = cv2.VideoCapture(temp_file.name)
|
86 |
+
|
87 |
+
if 'cap' in locals():
|
88 |
+
frame_size = (800, 480)
|
89 |
+
show_video = st.checkbox("Show video", value=True)
|
90 |
+
save_video = st.checkbox("Save video", value=False)
|
91 |
+
font = cv2.FONT_HERSHEY_DUPLEX
|
92 |
+
|
93 |
+
while True:
|
94 |
+
ret, frame = cap.read()
|
95 |
+
if ret:
|
96 |
+
frame = cv2.resize(frame, frame_size)
|
97 |
+
detections = detect_objects(frame)
|
98 |
+
display_frame = display_detections(frame, detections)
|
99 |
+
fps = 1 / (time.time() - start_time)
|
100 |
+
start_time = time.time()
|
101 |
+
cv2.putText(display_frame, f'FPS: {fps:.2f}', (10, 30), font, 1, (0, 255, 0), 2, cv2.LINE_AA)
|
102 |
+
if show_video:
|
103 |
+
stframe.image(display_frame, channels="BGR")
|
104 |
+
if save_video:
|
105 |
+
out.write(display_frame)
|
106 |
+
if cv2.waitKey(1) & 0xFF == ord('q'):
|
107 |
+
break
|
108 |
+
cap.release()
|
109 |
+
if save_video:
|
110 |
+
out.release()
|
model100e.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2a9d58c38bf2cf302bc1a4d75e42e16de1c9dabe3d7fe94d8b415a08d4e5857a
|
3 |
+
size 3906877
|
requirements.txt
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Pillow # PIL
|
2 |
+
opencv-python
|
3 |
+
torch
|
4 |
+
torchvision
|
5 |
+
numpy
|
6 |
+
tqdm
|
7 |
+
pandas
|
8 |
+
matplotlib
|