RMBG 2.0

🚀 BRIA Background Removal v2.0 Model Card

BRIA Background Removal v2.0 is a state-of-the-art model that effectively separates foreground from background in various image types. It is suitable for commercial use cases and offers high accuracy, efficiency, and versatility.

🚀 Quick Start

RMBG v2.0 is our new state-of-the-art background removal model that significantly improves upon RMBG v1.4. It's designed to effectively separate foreground from background across a range of categories and image types. Trained on a carefully selected dataset including general stock images, e-commerce, gaming, and advertising content, it's suitable for commercial use cases in enterprise content creation at scale. Its accuracy, efficiency, and versatility currently rival leading source-available models, making it ideal for scenarios where content safety, legally licensed datasets, and bias mitigation are crucial.

Developed by BRIA AI, RMBG v2.0 is available as a source-available model for non-commercial use.

Get Access

Bria RMBG2.0 is available wherever you build, whether as source-code and weights, ComfyUI nodes, or API endpoints.

• Purchase: For a commercial license, simply click Here.
• API Endpoint: Bria.ai, fal.ai
• ComfyUI: Use it in workflows

For more information, please visit our website.

Join our Discord community for more information, tutorials, tools, and to connect with other users!

CLICK HERE FOR A DEMO

✨ Features

• High Accuracy: Trained on a diverse dataset, it can accurately separate foreground from background in various image types.
• Versatility: Suitable for a wide range of commercial use cases, including enterprise content creation.
• Legal Compliance: Uses legally licensed datasets, ensuring content safety and bias mitigation.

📦 Installation

The model can be accessed in multiple ways:

• Source-code and weights: Available for non-commercial use.
• ComfyUI nodes: Use it in workflows
• API endpoints: Bria.ai, fal.ai

💻 Usage Examples

Basic Usage

from PIL import Image
import matplotlib.pyplot as plt
import torch
from torchvision import transforms
from transformers import AutoModelForImageSegmentation

model = AutoModelForImageSegmentation.from_pretrained('briaai/RMBG-2.0', trust_remote_code=True)
torch.set_float32_matmul_precision(['high', 'highest'][0])
model.to('cuda')
model.eval()

# Data settings
image_size = (1024, 1024)
transform_image = transforms.Compose([
    transforms.Resize(image_size),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])

image = Image.open(input_image_path)
input_images = transform_image(image).unsqueeze(0).to('cuda')

# Prediction
with torch.no_grad():
    preds = model(input_images)[-1].sigmoid().cpu()
pred = preds[0].squeeze()
pred_pil = transforms.ToPILImage()(pred)
mask = pred_pil.resize(image.size)
image.putalpha(mask)

image.save("no_bg_image.png")

📚 Documentation

Model Details

Model Description

Training data

The Bria-RMBG model was trained with over 15,000 high-quality, high-resolution, manually labeled (pixel-wise accuracy), fully licensed images. Our benchmark included balanced gender, balanced ethnicity, and people with different types of disabilities. For clarity, we provide our data distribution according to different categories, demonstrating our model’s versatility.

Distribution of images:

Qualitative Evaluation

Open source models comparison

Architecture

RMBG-2.0 is developed on the BiRefNet architecture enhanced with our proprietary dataset and training scheme. This training data significantly improves the model’s accuracy and effectiveness for the background-removal task.

If you use this model in your research, please cite:

@article{BiRefNet,
  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},
  year={2024}
}

Requirements

torch
torchvision
pillow
kornia
transformers

🔧 Technical Details

Model Architecture

The RMBG-2.0 model is built on the BiRefNet architecture. Our proprietary dataset and training scheme are used to enhance this architecture, which significantly improves the model’s accuracy and effectiveness for the background-removal task.

Training Dataset

The model is trained on a carefully selected dataset that includes general stock images, e-commerce, gaming, and advertising content. The dataset consists of over 15,000 high-quality, high-resolution, manually labeled (pixel-wise accuracy), fully licensed images. It also includes a balanced representation of gender, ethnicity, and people with different types of disabilities.

Model Output

The model output includes a single-channel 8-bit grayscale alpha matte, where each pixel value indicates the opacity level of the corresponding pixel in the original image. This non-binary output approach offers developers the flexibility to define custom thresholds for foreground-background separation, catering to varied use cases requirements and enhancing integration into complex pipelines.

📄 License

The model is released under the Creative Commons Attribution–Non-Commercial (CC BY-NC 4.0) license for non-commercial use. Commercial use is subject to a commercial agreement with BRIA. For more information, please visit here. To purchase a commercial license, simply click Here.