DEIM

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1999 stars · 212 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

DEIMv2 is a real-time object detection framework that extends the DEIM DETR family with DINOv3-pretrained and distilled backbones plus a Spatial Tuning Adapter. It ships eight model sizes from X down to Atto with pretrained weights, achieving state-of-the-art accuracy-efficiency trade-offs on COCO for GPU, edge, and mobile deployment.

Use cases

• train a real-time object detection model on custom data
• run state-of-the-art object detection on edge devices
• fine-tune DINOv3-backed detectors on COCO
• deploy lightweight object detection on mobile
• benchmark detection models against YOLO variants
• convert a single-scale foundation model backbone into multi-scale detection features

When to choose

• you need top accuracy-per-parameter real-time detection with pretrained weights
• you want DETR-style detection that converges fast
• you need model sizes spanning from mobile (Atto) to server (X)

When to avoid

• you need a permissively licensed model (custom DEIMv2 license)
• you need segmentation or pose estimation out of the box (see EdgeCrafter)
• you require a non-PyTorch deployment stack without conversion effort

Facets

library · maturity active

machine-learning computer-vision image-processing deep-learning computer-vision deep-learning machine-learning python cross-platform object-detection detr dinov3 real-time-detection detection-transformer model-zoo pretrained-weights coco gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Intellindust-AI-Lab/DEIMv2")