SAM3 - Semantic Segmentation Model

SAM3 is a semantic segmentation model deployed as a custom Docker container on HuggingFace Inference Endpoints.

🚀 Deployment

• GitHub Repository: https://github.com/logiroad/sam3
• Inference Endpoint: https://p6irm2x7y9mwp4l4.us-east-1.aws.endpoints.huggingface.cloud
• Docker Registry: sam3acr4hf.azurecr.io/sam3-hf:latest
• Model: facebook/sam3 (Sam3Model for static images)
• Hardware: NVIDIA A10G (24GB VRAM)

📊 Model Architecture

Built on Meta's SAM3 (Segment Anything Model 3) architecture for text-prompted semantic segmentation of static images.

🎯 Usage

import requests
import base64

# Read image
with open("image.jpg", "rb") as f:
    image_b64 = base64.b64encode(f.read()).decode()

# Call endpoint
response = requests.post(
    "https://p6irm2x7y9mwp4l4.us-east-1.aws.endpoints.huggingface.cloud",
    json={
        "inputs": image_b64,
        "parameters": {"classes": ["pothole", "asphalt"]}
    }
)

# Get results
masks = response.json()
for result in masks:
    print(f"Class: {result['label']}, Score: {result['score']}")

📦 Deployment

This model is deployed using a custom Docker image. See the GitHub repository for full documentation and deployment instructions.

📄 License

MIT License. This deployment uses Meta's SAM3 model - see the facebook/sam3 model card for model license information.

🔗 Resources

• Paper: SAM 3: Segment Anything with Concepts
• Full Documentation: GitHub Repository
• Endpoint Console: HuggingFace Endpoints