Spaces:
Sleeping
Sleeping
Add Files
Browse files- bg.png +0 -0
- main.py +59 -0
- model_0001999.pth +3 -0
- requirements.txt +9 -0
- util.py +103 -0
bg.png
ADDED
main.py
ADDED
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from detectron2.config import get_cfg
|
3 |
+
from detectron2.engine import DefaultPredictor
|
4 |
+
from detectron2 import model_zoo
|
5 |
+
from PIL import Image
|
6 |
+
import numpy as np
|
7 |
+
|
8 |
+
from util import visualize, set_background
|
9 |
+
|
10 |
+
|
11 |
+
set_background('bg.png')
|
12 |
+
|
13 |
+
|
14 |
+
# set title
|
15 |
+
st.title('Brain MRI tumor detection')
|
16 |
+
|
17 |
+
# set header
|
18 |
+
st.header('Please upload an image')
|
19 |
+
|
20 |
+
# upload file
|
21 |
+
file = st.file_uploader('', type=['png', 'jpg', 'jpeg'])
|
22 |
+
|
23 |
+
# load model
|
24 |
+
cfg = get_cfg()
|
25 |
+
cfg.merge_from_file(model_zoo.get_config_file('COCO-Detection/retinanet_R_101_FPN_3x.yaml'))
|
26 |
+
cfg.MODEL.WEIGHTS = 'model_0001999.pth'
|
27 |
+
cfg.MODEL.DEVICE = 'cpu'
|
28 |
+
|
29 |
+
predictor = DefaultPredictor(cfg)
|
30 |
+
|
31 |
+
# load image
|
32 |
+
if file:
|
33 |
+
image = Image.open(file).convert('RGB')
|
34 |
+
|
35 |
+
image_array = np.asarray(image)
|
36 |
+
|
37 |
+
# detect objects
|
38 |
+
outputs = predictor(image_array)
|
39 |
+
|
40 |
+
threshold = 0.5
|
41 |
+
|
42 |
+
# Display predictions
|
43 |
+
preds = outputs["instances"].pred_classes.tolist()
|
44 |
+
scores = outputs["instances"].scores.tolist()
|
45 |
+
bboxes = outputs["instances"].pred_boxes
|
46 |
+
|
47 |
+
bboxes_ = []
|
48 |
+
for j, bbox in enumerate(bboxes):
|
49 |
+
bbox = bbox.tolist()
|
50 |
+
|
51 |
+
score = scores[j]
|
52 |
+
pred = preds[j]
|
53 |
+
|
54 |
+
if score > threshold:
|
55 |
+
x1, y1, x2, y2 = [int(i) for i in bbox]
|
56 |
+
bboxes_.append([x1, y1, x2, y2])
|
57 |
+
|
58 |
+
# visualize
|
59 |
+
visualize(image, bboxes_)
|
model_0001999.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ea65ddcf276040cb0d20480482d5d8e8964deccfc18641ce583bf5ee99f97262
|
3 |
+
size 455037799
|
requirements.txt
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
streamlit==1.23.1
|
2 |
+
Pillow==9.5.0
|
3 |
+
numpy==1.24.3
|
4 |
+
torch==2.0.0
|
5 |
+
torchvision==0.15.1
|
6 |
+
opencv-python==4.6.0.66
|
7 |
+
matplotlib==3.5.3
|
8 |
+
plotly==5.15.0
|
9 |
+
git+https://github.com/facebookresearch/detectron2.git
|
util.py
ADDED
@@ -0,0 +1,103 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import base64
|
2 |
+
|
3 |
+
import plotly.graph_objects as go
|
4 |
+
import streamlit as st
|
5 |
+
|
6 |
+
|
7 |
+
def set_background(image_file):
|
8 |
+
"""
|
9 |
+
This function sets the background of a Streamlit app to an image specified by the given image file.
|
10 |
+
|
11 |
+
Parameters:
|
12 |
+
image_file (str): The path to the image file to be used as the background.
|
13 |
+
|
14 |
+
Returns:
|
15 |
+
None
|
16 |
+
"""
|
17 |
+
with open(image_file, "rb") as f:
|
18 |
+
img_data = f.read()
|
19 |
+
b64_encoded = base64.b64encode(img_data).decode()
|
20 |
+
style = f"""
|
21 |
+
<style>
|
22 |
+
.stApp {{
|
23 |
+
background-image: url(data:image/png;base64,{b64_encoded});
|
24 |
+
background-size: cover;
|
25 |
+
}}
|
26 |
+
</style>
|
27 |
+
"""
|
28 |
+
st.markdown(style, unsafe_allow_html=True)
|
29 |
+
|
30 |
+
|
31 |
+
def visualize(image, bboxes):
|
32 |
+
"""
|
33 |
+
Visualizes the image with bounding boxes using Plotly.
|
34 |
+
|
35 |
+
Args:
|
36 |
+
image: The input image.
|
37 |
+
bboxes (list): A list of bounding boxes in the format [x1, y1, x2, y2].
|
38 |
+
|
39 |
+
"""
|
40 |
+
# Get the width and height of the image
|
41 |
+
width, height = image.size
|
42 |
+
|
43 |
+
shapes = []
|
44 |
+
for bbox in bboxes:
|
45 |
+
x1, y1, x2, y2 = bbox
|
46 |
+
|
47 |
+
# Convert bounding box coordinates to the format expected by Plotly
|
48 |
+
shapes.append(dict(
|
49 |
+
type="rect",
|
50 |
+
x0=x1,
|
51 |
+
y0=height - y2,
|
52 |
+
x1=x2,
|
53 |
+
y1=height - y1,
|
54 |
+
line=dict(color='red', width=6),
|
55 |
+
))
|
56 |
+
|
57 |
+
fig = go.Figure()
|
58 |
+
|
59 |
+
# Add the image as a layout image
|
60 |
+
fig.update_layout(
|
61 |
+
images=[dict(
|
62 |
+
source=image,
|
63 |
+
xref="x",
|
64 |
+
yref="y",
|
65 |
+
x=0,
|
66 |
+
y=height,
|
67 |
+
sizex=width,
|
68 |
+
sizey=height,
|
69 |
+
sizing="stretch"
|
70 |
+
)]
|
71 |
+
)
|
72 |
+
|
73 |
+
# Set the axis ranges and disable axis labels
|
74 |
+
fig.update_xaxes(range=[0, width], showticklabels=False)
|
75 |
+
fig.update_yaxes(scaleanchor="x",
|
76 |
+
scaleratio=1,
|
77 |
+
range=[0, width], showticklabels=False)
|
78 |
+
|
79 |
+
fig.update_layout(
|
80 |
+
height=800,
|
81 |
+
updatemenus=[
|
82 |
+
dict(
|
83 |
+
direction='left',
|
84 |
+
pad=dict(r=10, t=10),
|
85 |
+
showactive=True,
|
86 |
+
x=0.11,
|
87 |
+
xanchor="left",
|
88 |
+
y=1.1,
|
89 |
+
yanchor="top",
|
90 |
+
type="buttons",
|
91 |
+
buttons=[
|
92 |
+
dict(label="Original",
|
93 |
+
method="relayout",
|
94 |
+
args=["shapes", []]),
|
95 |
+
dict(label="Detections",
|
96 |
+
method="relayout",
|
97 |
+
args=["shapes", shapes])
|
98 |
+
],
|
99 |
+
)
|
100 |
+
]
|
101 |
+
)
|
102 |
+
|
103 |
+
st.plotly_chart(fig)
|