BiSeNet (Bilateral Segmentation Network) is a state-of-the-art model for real-time semantic segmentation, initially proposed in the paper Bilateral Segmentation Network for Real-time Semantic Segmentation.

It combines two complementary paths:

  • Spatial Path: Captures high-resolution spatial information.
  • Context Path: Aggregates rich context information with a lightweight backbone.

The fusion of these paths ensures high segmentation accuracy with low computational cost, making it ideal for applications requiring real-time performance.

Downloads GitHub Repo stars GitHub Repository

🌟 Features

  • 🎯 Accurate Facial Parsing: Segments detailed facial features for precise analysis.
  • 🔄 ONNX Support: Torch-to-ONNX conversion for seamless deployment.
  • 🛠️ Enhanced Backbones: Flexible support for ResNet18 and ResNet34 models.
  • ⚡ High Efficiency: Optimized for real-time applications.

🖼️ Example Results

Input Images

ResNet34 Results

ResNet18 Results

🚀 Get Started

Clone the repository and explore the possibilities:

git clone https://github.com/yakhyo/face-parsing.git
cd face-parsing
pip install -r requirements.txt

Pre-trained weights are available for 🟢 ResNet18 and 🟠 ResNet34.

Visit the https://github.com/yakhyo/face-parsing for more details.