Feature: Added Enhance Finger Print Image

.gitignore

@@ -0,0 +1,5 @@
+*.jpg
+*.png
+__pycache__/
+venv/
+.env

.u2net/u2net.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d10d2f3bb75ae3b6d527c77944fc5e7dcd94b29809d47a739a7a728a912b491
+size 175997641

Dockerfile

@@ -4,6 +4,9 @@ FROM python:3.11-slim
 
 ENV PYTHONUNBUFFERED=1
 
+# rembg to use a local folder 
+ENV U2NET_HOME=/code/.u2net
+
 # Set the working directory
 WORKDIR /code
 
@@ -20,11 +23,15 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
+# Create necessary directories and set permissions
+RUN mkdir -p /code/model /code/.u2net && chmod -R 777 /code
+
 # Copy the rest of the application code
 COPY . .
 
-# Create a directory for the model if it doesn't exist (to match your app.py logic)
-RUN mkdir -p /code/model
+# --- PRE-DOWNLOAD MODEL ---
+# This "cooks" the 176MB model into your Docker layer so the Space starts instantly
+RUN python -c "from rembg import new_session; new_session('u2net')"
 
 # Expose the port Hugging Face uses
 EXPOSE 7860

app.py

@@ -6,13 +6,22 @@ start_import = time.time()
 os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'  # Hide everything but errors
 os.environ['CUDA_VISIBLE_DEVICES'] = '-1' # Force CPU mode immediately
 
-from flask import Flask, request, jsonify
+
+
+from flask import Flask, request, jsonify, Response
 import cv2
 import numpy as np
 import tensorflow as tf
 from tensorflow.keras.models import load_model
 from os.path import join, dirname
 
+
+from rembg import remove, new_session  # <--- Added rembg
+from PIL import Image
+
+
+# Set model path for rembg (helpful for Docker/HuggingFace)
+os.environ['U2NET_HOME'] = join(dirname(__file__), '.u2net')
 # Import your custom enhancement functions
 # Ensure 'enhancer.py' is in the same directory
 try:
@@ -33,18 +42,24 @@ MODEL_PATH = join(DIRNAME, 'model', '12_120_fp160.h5')
 # Global model variable
 model = None
 
+# --- Global Sessions ---
+# Load rembg session once to avoid reload overhead on every request
+rembg_session = new_session("u2net")
+siamese_model = None
+
 def load_and_warmup_model():
     global model
+    MODEL_PATH = join(dirname(__file__), 'model', '12_120_fp160.h5')
     t_start = time.time()
     try:
         print(f"Loading model from: {MODEL_PATH}")
-        model = load_model(MODEL_PATH, compile=False)
+        siamese_model = load_model(MODEL_PATH, compile=False)
         
         # Warm-up: Run a dummy inference so the first user request isn't slow
         print("Warming up model...")
         dummy_input = np.zeros((1, 160, 160, 1), dtype=np.float32)
         # Use predict_on_batch for faster single-sample inference
-        model.predict_on_batch([dummy_input, dummy_input])
+        siamese_model.predict_on_batch([dummy_input, dummy_input])
         
         print(f"Model loaded and warmed up in: {time.time() - t_start:.2f}s")
     except Exception as e:
@@ -53,6 +68,23 @@ def load_and_warmup_model():
 # Load model immediately on startup
 load_and_warmup_model()
 
+def remove_bg_from_cv2(cv2_img):
+    """
+    Converts CV2 image to PIL, removes background, 
+    and returns back to CV2 (BGR).
+    """
+    # Convert BGR to RGB for PIL
+    img_rgb = cv2.cvtColor(cv2_img, cv2.COLOR_BGR2RGB)
+    pil_img = Image.fromarray(img_rgb)
+    
+    # Remove background
+    output_pil = remove(pil_img, session=rembg_session)
+    
+    # Convert back to CV2 (and handle transparency)
+    # Note: rembg returns RGBA. We convert to BGR for your enhancer.
+    output_cv2 = cv2.cvtColor(np.array(output_pil), cv2.COLOR_RGBA2BGR)
+    return output_cv2
+
 def preprocess_fingerprint(image_bytes, image_type):
     """
     image_type: 'contactless' or 'contactbased'
@@ -156,6 +188,55 @@ def compare_fingerprints():
         }
     })
 
+@app.route('/enhance', methods=['POST'])
+def enhance_image_api():
+    """
+    API Endpoint: /enhance
+    Method: POST
+    Form-Data: 'image' (file)
+    Returns: JPEG image bytes of the enhanced fingerprint
+    """
+    # 1. Validation
+    if 'image' not in request.files:
+        return jsonify({"error": "No image file provided. Use key 'image'."}), 400
+
+    file = request.files['image']
+    
+    if file.filename == '':
+        return jsonify({"error": "No selected file"}), 400
+
+    try:
+        # 2. Decode Image
+        file_bytes = np.frombuffer(file.read(), np.uint8)
+        img = cv2.imdecode(file_bytes, cv2.IMREAD_COLOR)
+
+        if img is None:
+            return jsonify({"error": "Failed to decode image"}), 400
+
+        # 1. Background Removal
+        no_bg = remove_bg_from_cv2(img)
+
+        # 3. Apply Enhancement Pipeline (Contactless Logic: Basic -> Advanced)
+        # Note: basicEnhancing handles grayscale conversion internally if needed
+        step1 = basicEnhancing(no_bg)
+        
+        # advancedEnhancing expects the output from basicEnhancing
+        final_enhanced = advancedEnhancing(step1)
+
+        # 4. Encode Result to JPEG
+        # Using High Quality (95) to preserve ridge details
+        success, encoded_image = cv2.imencode('.jpg', final_enhanced, [int(cv2.IMWRITE_JPEG_QUALITY), 95])
+        
+        if not success:
+            return jsonify({"error": "Failed to encode resulting image"}), 500
+
+        # 5. Return Image Response
+        return Response(encoded_image.tobytes(), mimetype='image/jpeg')
+
+    except Exception as e:
+        print(f"Enhancement Error: {e}")
+        return jsonify({"error": str(e)}), 500
+
 if __name__ == '__main__':
     # Threaded=False is sometimes safer for TF, but Gunicorn handles the workers in prod.
     app.run(host='0.0.0.0', port=5000, debug=True)

enhancer.py

@@ -32,7 +32,7 @@ def advancedEnhancing(img):
 
 if __name__ == '__main__':
     # --- Execution Logic ---
-    image_path = r"C:\Users\SAGAR KESHAVE\Downloads\fore_tip.jpeg"
+    image_path = r"C:\SagarKV\sol9x\geekykant\contact-contactless-fingerprint\website\app\contactless_to_contactbased_fp_matching\finger_no_bg.png"
     # image_path = r"C:\SagarKV\sol9x\geekykant\contactless_2d_fingerprint_images\first_session\p247\p1.bmp"
     # image_path = r"C:\SagarKV\sol9x\geekykant\contact-based_fingerprints\second_session\13_3.jpg"
     original = cv2.imread(image_path)
@@ -69,8 +69,3 @@ if __name__ == '__main__':
         
         print(f"Comparison strip saved successfully as: {os.path.abspath(output_path)}")
 
-
-
-
-
-

requirements.txt

@@ -3,11 +3,13 @@ albucore==0.0.24
 albumentations==2.0.8
 annotated-types==0.7.0
 astunparse==1.6.3
+attrs==25.4.0
 blinker==1.9.0
 certifi==2026.1.4
 charset-normalizer==3.4.4
 click==8.3.1
 colorama==0.4.6
+coloredlogs==15.0.1
 fingerprint-feature-extractor==0.0.10
 fingerprint_enhancer==0.0.14
 Flask==3.1.2
@@ -18,40 +20,55 @@ google-pasta==0.2.0
 grpcio==1.76.0
 gunicorn==24.0.0
 h5py==3.15.1
+humanfriendly==10.0
 idna==3.11
 ImageIO==2.37.2
 itsdangerous==2.2.0
 Jinja2==3.1.6
 jsonpickle==4.1.1
+jsonschema==4.26.0
+jsonschema-specifications==2025.9.1
 keras==3.13.1
 lazy_loader==0.4
 libclang==18.1.1
+llvmlite==0.46.0
 Markdown==3.10.1
 markdown-it-py==4.0.0
 MarkupSafe==3.0.3
 mdurl==0.1.2
 ml_dtypes==0.5.4
+mpmath==1.3.0
 namex==0.1.0
 networkx==3.6.1
-numpy==2.4.1
+numba==0.63.1
+numpy==2.3.5
+onnxruntime==1.23.2
 opencv-python==4.13.0.90
 opencv-python-headless==4.13.0.90
 opt_einsum==3.4.0
 optree==0.18.0
 packaging==26.0
 pillow==12.1.0
+platformdirs==4.5.1
+pooch==1.8.2
 protobuf==6.33.4
 pydantic==2.12.5
 pydantic_core==2.41.5
 Pygments==2.19.2
+PyMatting==1.1.15
+pyreadline3==3.5.4
 PyYAML==6.0.3
+referencing==0.37.0
+rembg==2.0.72
 requests==2.32.5
 rich==14.2.0
+rpds-py==0.30.0
 scikit-image==0.26.0
 scipy==1.17.0
 simsimd==6.5.12
 six==1.17.0
 stringzilla==4.6.0
+sympy==1.14.0
 tensorboard==2.20.0
 tensorboard-data-server==0.7.2
 tensorflow==2.20.0