Spaces:
Sleeping
Sleeping
Update src/streamlit_app.py
Browse files- src/streamlit_app.py +11 -13
src/streamlit_app.py
CHANGED
@@ -4,12 +4,10 @@ import torch.nn as nn
|
|
4 |
import timm
|
5 |
import numpy as np
|
6 |
from PIL import Image
|
7 |
-
import requests
|
8 |
-
from io import BytesIO
|
9 |
import torchvision.transforms as T
|
10 |
-
import matplotlib.pyplot as plt
|
11 |
from huggingface_hub import hf_hub_download
|
12 |
|
|
|
13 |
class MobileViTSegmentation(nn.Module):
|
14 |
def __init__(self, encoder_name='mobilevit_s', pretrained=True):
|
15 |
super().__init__()
|
@@ -36,13 +34,12 @@ class MobileViTSegmentation(nn.Module):
|
|
36 |
# ========== Load Model ==========
|
37 |
@st.cache_resource
|
38 |
def load_model():
|
39 |
-
|
40 |
-
cache_dir = "/tmp/huggingface" # Writable path in HF Spaces
|
41 |
|
42 |
checkpoint_path = hf_hub_download(
|
43 |
repo_id="svsaurav95/ToothSegmentation",
|
44 |
filename="mobilevit_teeth_segmentation.pth",
|
45 |
-
cache_dir=cache_dir
|
46 |
)
|
47 |
|
48 |
model = MobileViTSegmentation(pretrained=False)
|
@@ -52,14 +49,15 @@ def load_model():
|
|
52 |
|
53 |
model = load_model()
|
54 |
|
55 |
-
# ========== Image
|
56 |
transform = T.Compose([
|
57 |
T.Resize((256, 256)),
|
58 |
T.ToTensor()
|
59 |
])
|
60 |
|
61 |
-
# ==========
|
62 |
-
st.
|
|
|
63 |
|
64 |
uploaded_file = st.file_uploader("Upload a mouth image with visible teeth", type=["jpg", "jpeg", "png"])
|
65 |
|
@@ -70,12 +68,12 @@ if uploaded_file:
|
|
70 |
with torch.no_grad():
|
71 |
pred_mask = model(input_tensor)[0, 0].numpy()
|
72 |
|
73 |
-
#
|
74 |
pred_mask = (pred_mask > 0.7).astype(np.uint8) * 255
|
75 |
pred_mask = Image.fromarray(pred_mask).resize(image.size)
|
76 |
|
77 |
-
# Create overlay
|
78 |
-
overlay = Image.new("RGBA", image.size, (0, 0, 255, 100))
|
79 |
base = image.convert("RGBA")
|
80 |
pred_mask_rgba = Image.new("L", image.size, 0)
|
81 |
pred_mask_rgba.paste(255, mask=pred_mask)
|
@@ -86,4 +84,4 @@ if uploaded_file:
|
|
86 |
with col1:
|
87 |
st.image(image, caption="Original Image", use_container_width=True)
|
88 |
with col2:
|
89 |
-
st.image(final, caption="Tooth Segmentation
|
|
|
4 |
import timm
|
5 |
import numpy as np
|
6 |
from PIL import Image
|
|
|
|
|
7 |
import torchvision.transforms as T
|
|
|
8 |
from huggingface_hub import hf_hub_download
|
9 |
|
10 |
+
# ========== Model Definition ==========
|
11 |
class MobileViTSegmentation(nn.Module):
|
12 |
def __init__(self, encoder_name='mobilevit_s', pretrained=True):
|
13 |
super().__init__()
|
|
|
34 |
# ========== Load Model ==========
|
35 |
@st.cache_resource
|
36 |
def load_model():
|
37 |
+
cache_dir = "/tmp/huggingface" # Safe writable directory in HF Spaces
|
|
|
38 |
|
39 |
checkpoint_path = hf_hub_download(
|
40 |
repo_id="svsaurav95/ToothSegmentation",
|
41 |
filename="mobilevit_teeth_segmentation.pth",
|
42 |
+
cache_dir=cache_dir
|
43 |
)
|
44 |
|
45 |
model = MobileViTSegmentation(pretrained=False)
|
|
|
49 |
|
50 |
model = load_model()
|
51 |
|
52 |
+
# ========== Image Preprocessing ==========
|
53 |
transform = T.Compose([
|
54 |
T.Resize((256, 256)),
|
55 |
T.ToTensor()
|
56 |
])
|
57 |
|
58 |
+
# ========== UI ==========
|
59 |
+
st.set_page_config(page_title="Tooth Segmentation", layout="wide")
|
60 |
+
st.title("🦷 Tooth Segmentation using MobileViT")
|
61 |
|
62 |
uploaded_file = st.file_uploader("Upload a mouth image with visible teeth", type=["jpg", "jpeg", "png"])
|
63 |
|
|
|
68 |
with torch.no_grad():
|
69 |
pred_mask = model(input_tensor)[0, 0].numpy()
|
70 |
|
71 |
+
# Threshold and resize to original
|
72 |
pred_mask = (pred_mask > 0.7).astype(np.uint8) * 255
|
73 |
pred_mask = Image.fromarray(pred_mask).resize(image.size)
|
74 |
|
75 |
+
# Create translucent blue overlay
|
76 |
+
overlay = Image.new("RGBA", image.size, (0, 0, 255, 100))
|
77 |
base = image.convert("RGBA")
|
78 |
pred_mask_rgba = Image.new("L", image.size, 0)
|
79 |
pred_mask_rgba.paste(255, mask=pred_mask)
|
|
|
84 |
with col1:
|
85 |
st.image(image, caption="Original Image", use_container_width=True)
|
86 |
with col2:
|
87 |
+
st.image(final, caption="Tooth Area Segmentation", use_container_width=True)
|