Welcome to the comprehensive repository designed to unleash the power of face recognition using OpenCV and TensorFlow on the NVIDIA Jetson Nano. This resourceful script capitalizes on advanced machine learning techniques, combining the robustness of OpenCV’s LBPHFaceRecognizer and the cutting-edge capabilities of TensorFlow models.
This script facilitates real-time face recognition through a webcam, making it a perfect tool for various applications from security systems to user authentication. Leveraging a pre-trained Haar Cascade model, the script adeptly detects faces in video streams. Post-detection, it employs dual recognition systems—OpenCV’s LBPH algorithm and TensorFlow’s neural network models—to accurately identify individuals.
- Dual Recognition System: Integrates the reliability of OpenCV’s LBPHFaceRecognizer and the advanced learning capabilities of TensorFlow's models.
- Real-Time Detection and Recognition: Seamlessly detects and identifies faces in real-time.
- Modularity: Easily adaptable for different use cases and datasets.
To run this script, ensure your environment is set up with the following:
- Python 3.6 or higher
- OpenCV-Python
- TensorFlow 2.x
You can install these dependencies via pip:
pip install numpy opencv-python tensorflow
Install git: How to install Git on any OS
Quick-Guide for git installation:
Installing Git Before running this project, you'll need Git installed on your system to manage version control. Here's how to install it for your operating system:
Download Link: https://git-scm.com/download/mac Installation: Download the installer and follow the on-screen prompts. Alternatively, you can use package managers like Homebrew (brew install git).
Use your system's package manager. Here are examples for common distributions:
Debian/Ubuntu: sudo apt install git
Fedora/CentOS/RHEL: sudo yum install git
Arch Linux: sudo pacman -S git
Download Link: https://git-scm.com/download/win Installation: Download the installer and follow the on-screen prompts.
Installing Other Dependencies
-
Clone the Repository: Clone this repository to your local machine to get started.
git clone https://github.com/VellVoid/Build-a-Face-Recognition-Machine-With-Jetson-Nano.git cd VellVoid/Build-a-Face-Recognition-Machine-With-Jetson-Nano -
Prepare the Dataset: Organize your dataset with folders labeled numerically for each individual.
Dataset Structure Create a root folder named dataset. Inside dataset, create subfolders for each person (e.g., "Alice", "Bob", "Cara").
Place multiple images of each person within their respective subfolders.
dataset/
Alice/
alice_image1.jpg
alice_image2.jpg
...
Bob/
bob_image1.jpg
bob_image2.jpg
...
...
-
Configure Your Environment: Set the environment variables to point to your dataset and model directories.
export DATASET_PATH=/path/to/your/datasetexport MODEL_PATH=/path/to/your/model
To run the face recognition script, navigate to the repository folder and execute:
python jetson_nano_face_recognition.py
- Face Detection: Uses a Haar Cascade Classifier to detect faces within the video frames.
- Recognition Process: Faces are analyzed using both LBPH and TensorFlow models to match against known identities.
- Result Display: Outputs the video with bounding boxes and labels indicating identified persons and their confidence levels.
- OpenCV Official Tutorials
- TensorFlow Full Course
- NVIDIA Developer Resources for Jetson Nano
- DIY Project on Instructables: Follow this detailed guide to build your own face recognition machine with Jetson Nano.
Special thanks to:
- The OpenCV and TensorFlow teams for their exceptional support and resources.
- NVIDIA for making high-performance computing accessible for developers and researchers.
Explore the possibilities of enhancing your projects with state-of-the-art face recognition technology using the Jetson Nano!
This project is open-sourced under the MIT License. See the LICENSE file for more details.