Upload 6 files

Dockerfile

@@ -0,0 +1,25 @@
+# ---- build stage ----
+FROM python:3.11-slim AS build
+
+WORKDIR /app
+COPY requirements.txt .
+RUN pip install --upgrade pip && \
+    pip wheel --no-deps -r requirements.txt -w /wheels
+
+# ---- runtime stage ----
+FROM python:3.11-slim
+
+ENV PYTHONUNBUFFERED=1
+WORKDIR /app
+
+# add wheels then install
+COPY --from=build /wheels /wheels
+RUN pip install --no-index --find-links=/wheels /wheels/*
+
+# copy source
+COPY app ./app
+COPY logs ./logs      # create if empty
+RUN mkdir -p temp_uploads
+
+EXPOSE 8000
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]

README.md

@@ -0,0 +1,3 @@
+# 🛡️ Fake-Logo Detector API
+
+Edge-friendly backend (FastAPI + TFLite) that spots *Real vs Fake* brand logos from a single photo and streams analytics.

labels.txt

@@ -0,0 +1,2 @@
+Genuine
+Fake

main.py

@@ -0,0 +1,48 @@
+from fastapi import FastAPI, File, UploadFile, Form
+from fastapi.responses import JSONResponse
+from fastapi.middleware.cors import CORSMiddleware
+from app.utils.predict import predict_logo
+import shutil
+import os
+import uuid
+
+app = FastAPI(title="Fake Logo Detector API")
+
+# Allow frontend calls (camera/gallery)
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+UPLOAD_DIR = "temp_uploads"
+os.makedirs(UPLOAD_DIR, exist_ok=True)
+
+@app.post("/scan-logo/")
+async def scan_logo(image: UploadFile = File(...), brand: str = Form("Unknown")):
+    try:
+        # Save the uploaded image temporarily
+        temp_filename = os.path.join(UPLOAD_DIR, f"{uuid.uuid4()}.jpg")
+        with open(temp_filename, "wb") as buffer:
+            shutil.copyfileobj(image.file, buffer)
+
+        # Call prediction logic
+        verdict, confidence = predict_logo(temp_filename)
+
+        # Delete temp file
+        os.remove(temp_filename)
+
+        return JSONResponse(content={
+            "brand": brand,
+            "verdict": verdict,
+            "confidence": round(confidence * 100, 2)
+        })
+
+    except Exception as e:
+        return JSONResponse(status_code=500, content={"error": str(e)})
+
+@app.get("/")
+async def root():
+    return {"message": "Fake Logo Detector API is running!"}

predict.py

@@ -0,0 +1,53 @@
+import numpy as np
+from PIL import Image
+import csv, os, datetime
+from tflite_runtime.interpreter import Interpreter   # use TensorFlow Lite runtime (tiny!)
+
+MODEL_PATH = os.path.join(os.path.dirname(_file_), "..", "model", "best_float32.tflite")
+LABELS_PATH = os.path.join(os.path.dirname(_file_), "..", "model", "labels.txt")
+LOG_PATH    = os.path.join(os.path.dirname(_file_), "..", "..", "logs")
+os.makedirs(LOG_PATH, exist_ok=True)
+LOG_FILE    = os.path.join(LOG_PATH, "detections.csv")
+
+# --- load model once ---------------------------------------------------------
+interpreter = Interpreter(model_path=MODEL_PATH)
+interpreter.allocate_tensors()
+
+input_details  = interpreter.get_input_details()
+output_details = interpreter.get_output_details()
+input_height, input_width = input_details[0]["shape"][1:3]
+
+# binary labels: idx 0 = Fake, idx 1 = Real
+with open(LABELS_PATH, "r") as f:
+    labels = [l.strip() for l in f.readlines()]
+
+# -----------------------------------------------------------------------------    
+def _preprocess(image_path: str) -> np.ndarray:
+    img = Image.open(image_path).convert("RGB").resize((input_width, input_height))
+    arr = np.asarray(img, dtype=np.float32) / 255.0          # normalize 0-1
+    arr = np.expand_dims(arr, axis=0)                        # add batch dim
+    return arr
+
+def _log(brand: str, verdict: str, conf: float):
+    is_new = not os.path.exists(LOG_FILE)
+    with open(LOG_FILE, "a", newline="") as f:
+        w = csv.writer(f)
+        if is_new:
+            w.writerow(["timestamp", "brand", "verdict", "confidence"])
+        w.writerow([datetime.datetime.now().isoformat(timespec="seconds"),
+                    brand, verdict, f"{conf:.4f}"])
+
+# -----------------------------------------------------------------------------    
+def predict_logo(image_path: str, brand: str = "Unknown"):
+    """Returns verdict ('Real'/'Fake') & confidence (0-1 float)."""
+    inp = _preprocess(image_path)
+    interpreter.set_tensor(input_details[0]["index"], inp)
+    interpreter.invoke()
+    output = interpreter.get_tensor(output_details[0]["index"])[0]  # shape (2,)
+    
+    conf_real = float(output[1])            # confidence for 'Real'
+    verdict   = "Real" if conf_real >= 0.5 else "Fake"
+    confidence = conf_real if verdict == "Real" else 1.0 - conf_real
+
+    _log(brand, verdict, confidence)
+    return verdict, confidences

requirements.txt

@@ -0,0 +1,6 @@
+fastapi==0.111.0
+uvicorn[standard]==0.29.0      # production ASGI server
+tflite-runtime==2.14.0         # 1-MB wheel — loads your .tflite model
+pillow==10.3.0                 # image handling
+python-multipart==0.0.9        # enables UploadFile in FastAPI
+numpy==1.26.4