Science Score: 26.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
  • codemeta.json file
    Found codemeta.json file
  • .zenodo.json file
    Found .zenodo.json file
  • DOI references
  • Academic publication links
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (8.0%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

Basic Info
  • Host: GitHub
  • Owner: GeorgePearse
  • License: apache-2.0
  • Language: Python
  • Default Branch: main
  • Size: 14.4 MB
Statistics
  • Stars: 1
  • Watchers: 1
  • Forks: 0
  • Open Issues: 20
  • Releases: 0
Created about 1 year ago · Last pushed about 1 year ago
Metadata Files
Readme Contributing License Citation

README.md

RF-DETR-MASK: Real-time Instance Segmentation with Transformer Magic

PyTorch Python COCO AP License

What is RF-DETR-MASK?

RF-DETR-MASK takes the lightning-fast object detection of RF-DETR and adds pixel-perfect instance segmentationthink of it as giving your AI superhuman vision that not only spots objects but traces their exact silhouettes!

Key Features

  • ** Real-time Performance**: Maintains RF-DETR's blazing speed while adding segmentation
  • ** High Accuracy**: Exceeds 60 AP on Microsoft COCO
  • ** Production Ready**: Optimized for edge deployment with ONNX export support
  • ** Multi-scale Features**: FPN-style architecture for capturing fine details
  • ** Modular Design**: Easy to extend and customize for your needs

Architecture Highlights

Built on the shoulders of giants, RF-DETR-MASK combines: - DINOv2 Backbone: State-of-the-art visual representations from Meta AI - Deformable Attention: Smart focus on relevant image regions - Lightweight Mask Head: Inspired by Facebook DETR - End-to-End Training: No complex post-processing needed!

** RF-DETR-MASK = Edge-friendly Performance + Pixel-precise Boundaries**

Owner

  • Name: George Pearse
  • Login: GeorgePearse
  • Kind: user
  • Location: London
  • Company: BinIt.ai

building mini educational ML and analytics apps and writing about them.

GitHub Events

Total
  • Issues event: 21
  • Watch event: 1
  • Delete event: 1
  • Issue comment event: 1
  • Push event: 113
  • Pull request event: 2
  • Create event: 6
Last Year
  • Issues event: 21
  • Watch event: 1
  • Delete event: 1
  • Issue comment event: 1
  • Push event: 113
  • Pull request event: 2
  • Create event: 6