BiRefNet_lite-2K Open-source Model - Free for High-resolution Background Removal and Mask Generation

Birefnet Lite 2K

Developed by ZhengPeng7

A bilateral reference framework for high-resolution binary image segmentation, focusing on background removal and mask generation tasks

Image Segmentation

Safetensors

#High-resolution image segmentation #Bilateral reference framework #2K resolution optimization

Downloads 3,400

Release Time : 9/24/2024

Model Overview

BiRefNet is a deep learning model for high-resolution binary image segmentation, particularly excelling in tasks such as background removal, mask generation, camouflaged object detection, and salient object detection. The model is trained at 2K resolution, making it suitable for scenarios requiring fine segmentation.

Model Features

High-resolution processing

Optimized specifically for 2K resolution (2560×1440), capable of handling fine segmentation tasks for high-resolution images

Multi-dataset training

Trained on multiple high-quality datasets including DIS5K, HRS10K, and UHRSD to enhance model generalization

Bilateral reference framework

Utilizes a unique bilateral reference architecture to improve segmentation accuracy and edge detail processing

Lightweight design

Offers a lite version that reduces computational resource requirements while maintaining performance

Model Capabilities

Background removal

Mask generation

Binary image segmentation

Camouflaged object detection

Salient object detection

Use Cases

Image editing

Product image background removal

Used for background removal and replacement of product images on e-commerce platforms

Achieves a maxFm score of 0.867 on the DIS-VD dataset

Portrait processing

Portrait matting

Used for precise segmentation of portraits in post-processing photography

Achieves a maxFm score of 0.993 on the TE-P3M-500-NP dataset

Medical image analysis

Medical image segmentation

Used for segmentation of specific tissues in CT/MRI images

🚀 Bilateral Reference for High-Resolution Dichotomous Image Segmentation

This project provides the official weights of BiRefNet_lite for high - resolution dichotomous image segmentation, supporting multiple image segmentation tasks.

🚀 Quick Start

This repo holds the official weights of BiRefNet_lite trained in 2K resolution.

✨ Features

Multiple Application Scenarios: Suitable for background removal, mask generation, dichotomous image segmentation, camouflaged object detection, and salient object detection.
High - Resolution Support: Trained with 2K resolution data, which can achieve better results in high - resolution image segmentation.

📦 Installation

No installation steps are provided in the original README, so this section is skipped.

💻 Usage Examples

No code examples are provided in the original README, so this section is skipped.

📚 Documentation

Training Sets

DIS5K (except DIS - VD)
HRS10K
UHRSD
P3M - 10k (except TE - P3M - 500 - NP)
TR - humans
AM - 2k
AIM - 500
Human - 2k (synthesized with BG - 20k)
Distinctions - 646 (synthesized with BG - 20k)
HIM2K
PPM - 100

HR samples selection

size_h, size_w = 1440, 2560
ratio = 0.8
h, w = image.shape[:2]
h >= size_h and w >= size_w or (h > size_h * ratio and w > size_w * ratio)

Validation Sets

DIS - VD
TE - P3M - 500 - NP

Performance

Dataset	Method	maxFm	wFmeasure	MAE	Smeasure	meanEm	HCE	maxEm	meanFm	adpEm	adpFm	mBA	maxBIoU	meanBIoU
DIS - VD	BiRefNet_lite - 2K - general--epoch_232	.867	.831	.045	.879	.919	952	.925	.858	.916	.847	.796	.750	.739
TE - P3M - 500 - NP	BiRefNet_lite - 2K - general--epoch_232	.993	.986	.009	.975	.986	.000	.993	.985	.833	.873	.825	.921	.891

Check more info

Check the main BiRefNet model repo for more info and how to use it:
https://huggingface.co/ZhengPeng7/BiRefNet/blob/main/README.md

⚠️ Important Note

Remember to set the resolution of input images to 2K (2560, 1440) for better results when using this model.

Also check the GitHub repo of BiRefNet for all things you may want:
https://github.com/ZhengPeng7/BiRefNet

🔧 Technical Details

No specific technical details (more than 50 words) are provided in the original README, so this section is skipped.

📄 License

The model is released under the MIT license.

<a href='LICENSE'><img src='https://img.shields.io/badge/License-MIT-yellow'></a>

Acknowledgement

Many thanks to @freepik for their generous support on GPU resources for training this model!

Citation

@article{zheng2024birefnet,
  title={Bilateral Reference for High-Resolution Dichotomous Image Segmentation},
  author={Zheng, Peng and Gao, Dehong and Fan, Deng-Ping and Liu, Li and Laaksonen, Jorma and Ouyang, Wanli and Sebe, Nicu},
  journal={CAAI Artificial Intelligence Research},
  volume = {3},
  pages = {9150038},
  year={2024}
}

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご