Spaces:
Sleeping
Sleeping
joshangngoching
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -4,10 +4,37 @@ from transformers import pipeline
|
|
4 |
import numpy as np
|
5 |
import cv2
|
6 |
import matplotlib.cm as cm
|
|
|
|
|
|
|
7 |
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
10 |
-
uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "png"])
|
11 |
|
12 |
def draw_masks_fromDict(image, results):
|
13 |
masked_image = image.copy()
|
@@ -26,16 +53,53 @@ def draw_masks_fromDict(image, results):
|
|
26 |
masked_image = masked_image.astype(np.uint8)
|
27 |
return cv2.addWeighted(image, 0.3, masked_image, 0.7, 0)
|
28 |
|
|
|
|
|
|
|
|
|
29 |
|
30 |
if uploaded_file is not None:
|
31 |
-
|
32 |
-
|
33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
|
35 |
-
|
36 |
-
st.json(segmentation_results)
|
37 |
|
38 |
-
|
|
|
|
|
|
|
|
|
|
|
39 |
|
40 |
-
|
41 |
-
|
|
|
|
4 |
import numpy as np
|
5 |
import cv2
|
6 |
import matplotlib.cm as cm
|
7 |
+
import time
|
8 |
+
import base64
|
9 |
+
from io import BytesIO
|
10 |
|
11 |
+
st.set_page_config(layout="wide")
|
12 |
+
|
13 |
+
with open("styles.css") as f:
|
14 |
+
st.markdown('<style>{}</style>'.format(f.read()), unsafe_allow_html=True)
|
15 |
+
|
16 |
+
|
17 |
+
st.markdown("<h1 class='title'>Segformer Semantic Segmentation</h1>", unsafe_allow_html=True)
|
18 |
+
st.markdown("""
|
19 |
+
<div class='text-center'>
|
20 |
+
This app uses the Segformer deep learning model to perform semantic segmentation on road images. The Transformer-based model is
|
21 |
+
trained on the CityScapes dataset which contains images of urban road scenes. Upload a
|
22 |
+
road scene and the app will return the image with semantic segmentation applied.
|
23 |
+
</div>
|
24 |
+
""", unsafe_allow_html=True)
|
25 |
+
|
26 |
+
group_members = ["Ang Ngo Ching, Josh Darren W.", "Bautista, Ryan Matthew M.", "Lacuesta, Angelo Giuseppe M.", "Reyes, Kenwin Hans", "Ting, Sidney Mitchell O."]
|
27 |
+
|
28 |
+
|
29 |
+
# model_versions = ["b1", "b2", "b3", "b4", "b5"]
|
30 |
+
# selected_model_version = st.selectbox("Select a model version:", model_versions)
|
31 |
+
|
32 |
+
|
33 |
+
semantic_segmentation = pipeline("image-segmentation", f"nvidia/segformer-b1-finetuned-cityscapes-1024-1024")
|
34 |
+
|
35 |
+
new_file_uploaded = False
|
36 |
+
uploaded_file = st.file_uploader("", type=["jpg", "png"])
|
37 |
|
|
|
38 |
|
39 |
def draw_masks_fromDict(image, results):
|
40 |
masked_image = image.copy()
|
|
|
53 |
masked_image = masked_image.astype(np.uint8)
|
54 |
return cv2.addWeighted(image, 0.3, masked_image, 0.7, 0)
|
55 |
|
56 |
+
col1, col2 = st.columns(2)
|
57 |
+
|
58 |
+
if "uploaded_file" not in st.session_state:
|
59 |
+
st.session_state.uploaded_file = None
|
60 |
|
61 |
if uploaded_file is not None:
|
62 |
+
st.session_state.uploaded_file = uploaded_file
|
63 |
+
|
64 |
+
if st.session_state.uploaded_file is not None:
|
65 |
+
image = Image.open(st.session_state.uploaded_file)
|
66 |
+
col1, col2 = st.columns(2)
|
67 |
+
|
68 |
+
with col1:
|
69 |
+
st.image(image, caption='Uploaded Image.', use_column_width=True)
|
70 |
+
|
71 |
+
|
72 |
+
while True:
|
73 |
+
with st.spinner('Processing...'):
|
74 |
+
segmentation_results = semantic_segmentation(image)
|
75 |
+
image_with_masks = draw_masks_fromDict(np.array(image), segmentation_results)
|
76 |
+
image_with_masks_pil = Image.fromarray(image_with_masks, 'RGB')
|
77 |
+
|
78 |
+
with col2:
|
79 |
+
st.image(image_with_masks_pil, caption='Segmented Image.', use_column_width=True)
|
80 |
+
|
81 |
+
buffered = BytesIO()
|
82 |
+
image_with_masks_pil.save(buffered, format="PNG")
|
83 |
+
img_str = base64.b64encode(buffered.getvalue()).decode()
|
84 |
+
href = f'<a href="data:file/png;base64,{img_str}" download="segmented_{st.session_state.uploaded_file.name}">Download Segmented Image</a>'
|
85 |
+
st.markdown(href, unsafe_allow_html=True)
|
86 |
+
|
87 |
+
new_file_uploaded = False
|
88 |
+
|
89 |
+
while not new_file_uploaded:
|
90 |
+
time.sleep(1)
|
91 |
+
|
92 |
+
|
93 |
|
94 |
+
pdf_url = "https://arxiv.org/pdf/2105.15203.pdf"
|
|
|
95 |
|
96 |
+
st.markdown("""
|
97 |
+
<h3 class='text-center'>
|
98 |
+
Read more about the paper below👇
|
99 |
+
</h5>
|
100 |
+
""", unsafe_allow_html=True)
|
101 |
+
st.markdown(f'<iframe class="pdf" src={pdf_url}></iframe>', unsafe_allow_html=True)
|
102 |
|
103 |
+
st.markdown("Group Members:")
|
104 |
+
for member in group_members:
|
105 |
+
st.markdown("- " + member)
|