Skip to content

UniFace: A Unified Face Analysis Library for Python | Detection, alignment, landmarks, face-mesh, recognition, parsing, gaze, attributes and anti-spoofing under one API.

7.0 KB
Raw Blame
{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Face Verification: One-to-One Face Comparison\n",
    "\n",
    "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/yakhyo/uniface/blob/main/examples/03_face_verification.ipynb) [![Open in Kaggle](https://kaggle.com/static/images/open-in-kaggle.svg)](https://www.kaggle.com/code/yakhyokhuja/face-verification-one-to-one-face-comparison)\n",
    "\n",
    "Compare two face images and decide whether they show the same person.\n",
    "\n",
    "<img src=\"https://raw.githubusercontent.com/yakhyo/uniface/main/assets/demo/verification.jpg\" width=\"100%\">\n",
    "\n",
    "---\n",
    "\n",
    "**UniFace** is a lightweight, production-ready Python library for face detection, recognition, tracking, landmark analysis, face parsing, gaze estimation, and face attributes.\n",
    "\n",
    "GitHub: [github.com/yakhyo/uniface](https://github.com/yakhyo/uniface) | Docs: [yakhyo.github.io/uniface](https://yakhyo.github.io/uniface)\n",
    "\n",
    "## Setup\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%pip install -q \"uniface[cpu]\"\n",
    "\n",
    "# Clone repo for assets (Colab / Kaggle only)\n",
    "import os\n",
    "if any(k in os.environ for k in ('COLAB_GPU', 'COLAB_RELEASE_TAG', 'KAGGLE_KERNEL_RUN_TYPE')):\n",
    "    if not os.path.exists('uniface'):\n",
    "        !git clone --depth 1 https://github.com/yakhyo/uniface.git\n",
    "    os.chdir('uniface/examples')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import uniface\n",
    "from uniface.analyzer import FaceAnalyzer\n",
    "from uniface.detection import RetinaFace\n",
    "from uniface.recognition import ArcFace\n",
    "\n",
    "print(uniface.__version__)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We need detection and recognition models for face verification.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "analyzer = FaceAnalyzer(\n",
    "    detector=RetinaFace(confidence_threshold=0.5),\n",
    "    recognizer=ArcFace()\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Load Two Faces\n",
    "\n",
    "Two photographs of the same person, fourteen years apart."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "image_path1 = '../assets/source/verify_now_2010.jpg'\n",
    "image_path2 = '../assets/source/verify_now_2024.jpg'\n",
    "\n",
    "image1 = cv2.imread(image_path1)\n",
    "image2 = cv2.imread(image_path2)\n",
    "\n",
    "# Analyze faces\n",
    "faces1 = analyzer.analyze(image1)\n",
    "faces2 = analyzer.analyze(image2)\n",
    "\n",
    "print(f'Detected {len(faces1)} and {len(faces2)} faces')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Look at Both Images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, axes = plt.subplots(1, 2, figsize=(10, 5))\n",
    "\n",
    "axes[0].imshow(cv2.cvtColor(image1, cv2.COLOR_BGR2RGB))\n",
    "axes[0].set_title('Image 1')\n",
    "axes[0].axis('off')\n",
    "\n",
    "axes[1].imshow(cv2.cvtColor(image2, cv2.COLOR_BGR2RGB))\n",
    "axes[1].set_title('Image 2')\n",
    "axes[1].axis('off')\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Compute Similarity\n",
    "\n",
    "`compute_similarity()` is a dot product between two normalised embeddings, so it returns\n",
    "the cosine similarity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "if faces1 and faces2:\n",
    "    face1 = faces1[0]\n",
    "    face2 = faces2[0]\n",
    "\n",
    "    similarity = face1.compute_similarity(face2)\n",
    "    print(f'Similarity: {similarity:.4f}')\n",
    "else:\n",
    "    print('Error: Could not detect faces')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Decide With a Threshold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "THRESHOLD = 0.4  # the library default, matching FAISS.search\n",
    "\n",
    "if faces1 and faces2:\n",
    "    is_same_person = similarity > THRESHOLD\n",
    "\n",
    "    print(f'Similarity: {similarity:.4f}')\n",
    "    print(f'Threshold: {THRESHOLD}')\n",
    "    print(f'Result: {\"Same person\" if is_same_person else \"Different people\"}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Compare Several Pairs\n",
    "\n",
    "The same person against themselves, then each of them against a different person."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "image_pairs = [\n",
    "    ('../assets/source/verify_now_2010.jpg', '../assets/source/verify_now_2014.jpg'),\n",
    "    ('../assets/source/verify_now_2014.jpg', '../assets/source/verify_now_2024.jpg'),\n",
    "    ('../assets/source/verify_now_2010.jpg', '../assets/source/verify_now_2024.jpg'),\n",
    "    ('../assets/source/verify_now_2024.jpg', '../assets/source/age_adult.jpg'),\n",
    "]\n",
    "\n",
    "print('Comparing multiple pairs:')\n",
    "for img1_path, img2_path in image_pairs:\n",
    "    img1 = cv2.imread(img1_path)\n",
    "    img2 = cv2.imread(img2_path)\n",
    "\n",
    "    faces_a = analyzer.analyze(img1)\n",
    "    faces_b = analyzer.analyze(img2)\n",
    "\n",
    "    if faces_a and faces_b:\n",
    "        sim = faces_a[0].compute_similarity(faces_b[0])\n",
    "\n",
    "        img1_name = img1_path.split('/')[-1]\n",
    "        img2_name = img2_path.split('/')[-1]\n",
    "\n",
    "        print(f'{img1_name} vs {img2_name}: {sim:.4f}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Notes\n",
    "\n",
    "- 0.4 is the library default, the same value `FAISS.search` uses.\n",
    "- No single threshold fits every dataset. Long time gaps, low resolution and scanned prints all\n",
    "  pull genuine scores down, so a real match can land below 0.6. Measure a few known pairs from\n",
    "  your own images before choosing a number."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.13.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}