Spaces:
Running
Running
from fastapi import FastAPI, File, UploadFile, Form, HTTPException | |
from fastapi.responses import JSONResponse, HTMLResponse | |
import gradio as gr | |
from deepface import DeepFace | |
import os | |
from threading import Thread | |
import asyncio | |
os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE" | |
# FastAPI instance | |
app = FastAPI() | |
# Gradio Interface Function | |
def face_verification_uii(img1, img2, dist="cosine", model="Facenet", detector="ssd"): | |
""" | |
Gradio function for face verification | |
""" | |
try: | |
result = DeepFace.verify( | |
img1_path=img1, | |
img2_path=img2, | |
distance_metric=dist, | |
model_name=model, | |
detector_backend=detector, | |
enforce_detection=False, | |
) | |
return { | |
"verified": result["verified"], | |
"distance": result["distance"], | |
"threshold": result["threshold"], | |
"model": result["model"], | |
"detector_backend": result["detector_backend"], | |
"similarity_metric": result["similarity_metric"], | |
} | |
except Exception as e: | |
return {"error": str(e)} | |
# FastAPI Endpoint | |
async def gradio_ui(): | |
html_content = """ | |
<html> | |
<head> | |
<title>Gradio UI</title> | |
</head> | |
<body> | |
<iframe src="http://localhost:7861" width="100%" height="100%" frameborder="0"></iframe> | |
</body> | |
</html> | |
""" | |
return HTMLResponse(content=html_content) | |
async def face_verification( | |
img1: UploadFile = File(...), | |
img2: UploadFile = File(...), | |
dist: str = Form("cosine"), | |
model: str = Form("Facenet"), | |
detector: str = Form("ssd") | |
): | |
""" | |
Endpoint to verify if two images belong to the same person. | |
""" | |
try: | |
# Ensure uploads directory exists | |
if not os.path.exists("uploads"): | |
os.makedirs("uploads") | |
# Save uploaded images to disk | |
img1_path = os.path.join("uploads", img1.filename) | |
img2_path = os.path.join("uploads", img2.filename) | |
with open(img1_path, "wb") as f: | |
f.write(await img1.read()) | |
with open(img2_path, "wb") as f: | |
f.write(await img2.read()) | |
# Run DeepFace verification | |
result = DeepFace.verify( | |
img1_path=img1_path, | |
img2_path=img2_path, | |
distance_metric=dist, | |
model_name=model, | |
detector_backend=detector, | |
enforce_detection=False, | |
) | |
# Delete uploaded images after processing | |
os.remove(img1_path) | |
os.remove(img2_path) | |
# Return verification results | |
return { | |
"verified": result["verified"], | |
"distance": result["distance"], | |
"threshold": result["threshold"], | |
"model": result["model"], | |
"detector_backend": result["detector_backend"], | |
"similarity_metric": result["similarity_metric"] | |
} | |
except Exception as e: | |
raise HTTPException(status_code=500, detail=str(e)) | |
def run_gradio_ui(): | |
""" | |
Function to run Gradio in a separate thread | |
""" | |
# Create and set an event loop for this thread | |
loop = asyncio.new_event_loop() | |
asyncio.set_event_loop(loop) | |
def face_verification_ui(img1, img2, dist, model, detector): | |
result = face_verification_uii(img1, img2, dist, model, detector) | |
return result | |
with gr.Blocks() as demo: | |
img1 = gr.Image(type="filepath", label="Image 1") | |
img2 = gr.Image(type="filepath", label="Image 2") | |
dist = gr.Dropdown(choices=["cosine", "euclidean", "euclidean_l2"], label="Distance Metric", value="cosine") | |
model = gr.Dropdown(choices=["VGG-Face", "Facenet", "Facenet512", "ArcFace"], label="Model", value="Facenet") | |
detector = gr.Dropdown(choices=["opencv", "ssd", "mtcnn", "retinaface", "mediapipe"], label="Detector", value="ssd") | |
btn = gr.Button("Verify") | |
output = gr.Textbox() | |
btn.click(face_verification_ui, inputs=[img1, img2, dist, model, detector], outputs=output) | |
demo.launch(server_name="0.0.0.0", server_port=7861, show_api=False) | |
# FastAPI Startup Event | |
# FastAPI Startup Event | |
def startup_event(): | |
""" | |
Start Gradio UI in a separate thread | |
""" | |
thread = Thread(target=run_gradio_ui) | |
thread.start() | |
# Running Both Servers | |
if __name__ == "__main__": | |
import uvicorn | |
uvicorn.run(app, host="0.0.0.0", port=7860, reload=True) |