Update app.py

app.py

@@ -1,63 +1,63 @@
-import gradio as gr
+# app.py
+from fastapi import FastAPI, File, UploadFile
+from fastapi.middleware.cors import CORSMiddleware
+from ultralytics import YOLO
+import uvicorn
 import numpy as np
 from PIL import Image
+import io
 
-from inference_utils import create_model, inference
+app = FastAPI()
 
-# --------- DeepCrack 옵션 구성 ---------
-class Opt:
-    # 기본값
-    checkpoints_dir = "./checkpoints"
-    name = "deepcrack"
-    gpu_ids = []
-    isTrain = False
-
-    input_nc = 3
-    num_classes = 1
-    ngf = 64
-
-    norm = "instance"
-    init_type = "normal"
-    init_gain = 0.02
-
-    display_sides = False
-    loss_mode = "bce"
-    lr = 0.001
-
-
-# --------- 모델 로드 ---------
-print("🔥 Loading DeepCrack model...")
-opt = Opt()
-model = create_model(opt, cp_path="pretrained_net_G.pth")
-print("🔥 DeepCrack model loaded!")
-
-
-# --------- 예측 함수 ---------
-def predict(img: Image.Image):
-    output_img, confidence = inference(model, img)
+# CORS 활성화 (ConcreteAI 웹과 연결)
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
 
-    has_crack = confidence > 0.5
-    label = "crack" if has_crack else "normal"
+# ---- YOLOv8 모델 로드 ----
+print("🔵 Loading YOLOv8 crack model...")
+model = YOLO("keremberke/yolov8n-concrete-crack")
+print("✅ Model loaded!")
+
+@app.post("/predict")
+async def predict(img: UploadFile = File(...)):
+    # 이미지 읽기
+    image_bytes = await img.read()
+    image = Image.open(io.BytesIO(image_bytes)).convert("RGB")
+    np_img = np.array(image)
+
+    # YOLOv8 추론
+    results = model(np_img)[0]
+
+    # 박스 리스트
+    detections = results.boxes
+
+    if detections is None or len(detections) == 0:
+        # 균열 없음
+        return {
+            "data": [
+                {
+                    "label": "normal",
+                    "confidence": 1.0
+                }
+            ]
+        }
+
+    # 가장 높은 confidence 선택
+    max_conf = float(max(d.conf[0].item() for d in detections))
 
     return {
         "data": [
             {
-                "label": label,
-                "confidence": float(confidence)
+                "label": "crack",
+                "confidence": max_conf
             }
         ]
     }
 
-
-# --------- Gradio API 인터페이스 ---------
-demo = gr.Interface(
-    fn=predict,
-    inputs=gr.Image(type="pil"),
-    outputs=gr.JSON(label="Detection Result"),
-    title="DeepCrack — Concrete Crack Detection",
-    description="딥러닝 기반 콘크리트 균열 segmentation 모델 DeepCrack",
-    flagging_mode="never"
-)
-
 if __name__ == "__main__":
-    demo.launch()
+    uvicorn.run(app, host="0.0.0.0", port=7860)