|
|
import gradio as gr |
|
|
import cv2 |
|
|
import torch |
|
|
|
|
|
|
|
|
model = torch.hub.load('models', 'custom', 'models/100epoch.pt', force_reload=True, source='local', trust_repo=True) |
|
|
model.eval() |
|
|
|
|
|
|
|
|
label_mapping = { |
|
|
'A': 'Adenocarcinoma', |
|
|
'B': 'Small Cell Carcinoma', |
|
|
'E': 'Large Cell Carcinoma', |
|
|
'G': 'Squamous Cell Carcinoma' |
|
|
} |
|
|
|
|
|
def process_image(input_image): |
|
|
|
|
|
results = model(input_image) |
|
|
img = results.render()[0] |
|
|
if results.pred is not None and len(results.pred[0]) > 0: |
|
|
detection = results.pred[0] |
|
|
class_index = int(detection[0, -1]) |
|
|
print(class_index) |
|
|
class_label = model.names[class_index] |
|
|
mapped_label = label_mapping.get(class_label, "Unknown") |
|
|
else: |
|
|
mapped_label = "No cancer detected" |
|
|
|
|
|
return img, mapped_label |
|
|
|
|
|
|
|
|
iface = gr.Interface( |
|
|
fn=process_image, |
|
|
inputs=gr.components.Image(type='pil', label="Input Image").style(height=280), |
|
|
outputs=[gr.components.Image(type='pil', label="Processed Image").style(height=280), gr.components.Textbox(label="Detected Cancer Type")], |
|
|
live=True, |
|
|
title="Lung Cancer Detector ⚕️", |
|
|
description="The AI model was trained to detect the following types of lung cancer:\n" |
|
|
"1. Adenocarcinoma (A)\n" |
|
|
"2. Small Cell Carcinoma (B)\n" |
|
|
"3. Large Cell Carcinoma (E)\n" |
|
|
"4. Squamous Cell Carcinoma (G)\n\n" |
|
|
"How to Use :\n" |
|
|
"1. Upload a CT scan image of a patient's lungs.\n" |
|
|
"2. The app will display the predicted type of lung cancer.", |
|
|
theme=gr.themes.Monochrome(font=[gr.themes.GoogleFont("Noto Serif"), "Preahvihear", "sans-serif"]) |
|
|
|
|
|
) |
|
|
|
|
|
if __name__ == '__main__': |
|
|
iface.launch() |
