Update app.py
Browse files
app.py
CHANGED
@@ -1,10 +1,10 @@
|
|
1 |
import torch
|
2 |
-
import torch
|
3 |
import torch.nn as nn
|
4 |
from transformers import ViTImageProcessor, ViTModel, BertTokenizerFast, BertModel
|
5 |
from PIL import Image
|
6 |
import gradio as gr
|
7 |
|
|
|
8 |
class VisionLanguageModel(nn.Module):
|
9 |
def __init__(self):
|
10 |
super(VisionLanguageModel, self).__init__()
|
@@ -33,7 +33,6 @@ class VisionLanguageModel(nn.Module):
|
|
33 |
logits = self.classifier(combined_features)
|
34 |
return logits
|
35 |
|
36 |
-
# Load the model checkpoint with safer loading
|
37 |
model = VisionLanguageModel()
|
38 |
model.load_state_dict(torch.load('best_model.pth', map_location=torch.device('cpu'), weights_only=True))
|
39 |
model.eval()
|
@@ -42,10 +41,7 @@ tokenizer = BertTokenizerFast.from_pretrained('bert-base-uncased')
|
|
42 |
feature_extractor = ViTImageProcessor.from_pretrained('google/vit-base-patch16-224-in21k')
|
43 |
|
44 |
def predict(image, text_input):
|
45 |
-
# Preprocess the image
|
46 |
image = feature_extractor(images=image, return_tensors="pt").pixel_values
|
47 |
-
|
48 |
-
# Preprocess the text
|
49 |
encoding = tokenizer(
|
50 |
text_input,
|
51 |
add_special_tokens=True,
|
@@ -54,8 +50,6 @@ def predict(image, text_input):
|
|
54 |
truncation=True,
|
55 |
return_tensors='pt'
|
56 |
)
|
57 |
-
|
58 |
-
# Make a prediction
|
59 |
with torch.no_grad():
|
60 |
outputs = model(
|
61 |
input_ids=encoding['input_ids'],
|
@@ -63,19 +57,44 @@ def predict(image, text_input):
|
|
63 |
pixel_values=image
|
64 |
)
|
65 |
_, prediction = torch.max(outputs, dim=1)
|
66 |
-
return
|
67 |
-
|
68 |
-
#
|
69 |
-
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
|
74 |
-
|
75 |
-
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
81 |
|
|
|
1 |
import torch
|
|
|
2 |
import torch.nn as nn
|
3 |
from transformers import ViTImageProcessor, ViTModel, BertTokenizerFast, BertModel
|
4 |
from PIL import Image
|
5 |
import gradio as gr
|
6 |
|
7 |
+
# Model definition and setup
|
8 |
class VisionLanguageModel(nn.Module):
|
9 |
def __init__(self):
|
10 |
super(VisionLanguageModel, self).__init__()
|
|
|
33 |
logits = self.classifier(combined_features)
|
34 |
return logits
|
35 |
|
|
|
36 |
model = VisionLanguageModel()
|
37 |
model.load_state_dict(torch.load('best_model.pth', map_location=torch.device('cpu'), weights_only=True))
|
38 |
model.eval()
|
|
|
41 |
feature_extractor = ViTImageProcessor.from_pretrained('google/vit-base-patch16-224-in21k')
|
42 |
|
43 |
def predict(image, text_input):
|
|
|
44 |
image = feature_extractor(images=image, return_tensors="pt").pixel_values
|
|
|
|
|
45 |
encoding = tokenizer(
|
46 |
text_input,
|
47 |
add_special_tokens=True,
|
|
|
50 |
truncation=True,
|
51 |
return_tensors='pt'
|
52 |
)
|
|
|
|
|
53 |
with torch.no_grad():
|
54 |
outputs = model(
|
55 |
input_ids=encoding['input_ids'],
|
|
|
57 |
pixel_values=image
|
58 |
)
|
59 |
_, prediction = torch.max(outputs, dim=1)
|
60 |
+
return prediction.item() # 1 for Malignant, 0 for Benign
|
61 |
+
|
62 |
+
# Enhanced UI with color-coded prediction display
|
63 |
+
with gr.Blocks(css="""
|
64 |
+
.benign {background-color: white; border: 1px solid lightgray; padding: 10px; border-radius: 5px;}
|
65 |
+
.malignant {background-color: white; border: 1px solid lightgray; padding: 10px; border-radius: 5px;}
|
66 |
+
.benign.correct {background-color: lightgreen;}
|
67 |
+
.malignant.correct {background-color: lightgreen;}
|
68 |
+
""") as demo:
|
69 |
+
gr.Markdown(
|
70 |
+
"""
|
71 |
+
# 🩺 SKIN LESION CLASSIFICATION
|
72 |
+
Upload an image of a skin lesion and provide clinical details to get a prediction of benign or malignant.
|
73 |
+
"""
|
74 |
+
)
|
75 |
+
|
76 |
+
with gr.Row():
|
77 |
+
with gr.Column(scale=1):
|
78 |
+
image_input = gr.Image(type="pil", label="Upload Skin Lesion Image")
|
79 |
+
text_input = gr.Textbox(label="Clinical Information (e.g., patient age, symptoms)")
|
80 |
+
|
81 |
+
with gr.Column(scale=1):
|
82 |
+
benign_output = gr.HTML("<div class='benign'>Benign</div>")
|
83 |
+
malignant_output = gr.HTML("<div class='malignant'>Malignant</div>")
|
84 |
+
gr.Markdown("## Example:")
|
85 |
+
example_image = gr.Image(value="skin_cancer_detection/Unknown-4.png") # Provide path to an example image
|
86 |
+
example_text = gr.Textbox(value="consistent with resolving/involuting keratoacanthoma 67", interactive=False)
|
87 |
+
|
88 |
+
def display_prediction(image, text_input):
|
89 |
+
prediction = predict(image, text_input)
|
90 |
+
benign_html = "<div class='benign{}'>Benign</div>".format(" correct" if prediction == 0 else "")
|
91 |
+
malignant_html = "<div class='malignant{}'>Malignant</div>".format(" correct" if prediction == 1 else "")
|
92 |
+
return benign_html, malignant_html
|
93 |
+
|
94 |
+
# Submit button and prediction outputs
|
95 |
+
submit_btn = gr.Button("Get Prediction")
|
96 |
+
submit_btn.click(display_prediction, inputs=[image_input, text_input], outputs=[benign_output, malignant_output])
|
97 |
+
|
98 |
+
demo.launch()
|
99 |
+
|
100 |
|