Pyyan / Compare / Grounding DINO vs RF-DETR vs YOLOv12 vs RTMDet
Grounding DINO vs RF-DETR vs YOLOv12 vs RTMDet
4 of 5Object Detection · verified 13 Aug 2026
| Specification | Grounding DINO | RF-DETR | YOLOv12 | RTMDet |
|---|---|---|---|---|
| Summary | Detects whatever you describe in words. | First real-time detector past 60 mAP on COCO. | Attention-centric YOLO, still widely deployed. | Pure throughput, and an MIT licence. |
| COCO mAP | ~52 zero-shot | 60.5 | ~55 | ~52 |
| Speed | Moderate | Real time | Very high | 300+ |
| Licence | Apache 2.0 | Apache 2.0 | AGPL-3.0 | MIT |
| Family | Open vocabulary | Transformer, set prediction | Single stage | Single stage |
| Open vocabulary | Yes | No | No | No |
| Released | 2023 | 2025 | 2025 | 2022 |
| Category | Object Detection | Object Detection | Object Detection | Object Detection |
| Official | IDEA Research ↗ | Roboflow ↗ | Ultralytics ↗ | OpenMMLab ↗ |
Highlighted rows are where these differ.
Grounding DINO
- Text prompt in, boxes out
- Fine-tuning still beats it on specialised objects
Best for classes you have no data for.
RF-DETR
- Leads the RF100-VL domain-transfer benchmark
- DINOv2 backbone, markedly better on occluded objects
- Apache 2.0, which YOLO is not
Best for most custom detection tasks.
YOLOv12
- Attention added to the classic single-stage design
- Same licensing consideration as YOLO26
Best for teams already on the YOLO toolchain.
RTMDet
- Wins on raw speed where accuracy is sufficient
- MIT, so no derivative-work obligation
Best for high frame-rate video.