Pyyan / Compare / RTMDet vs YOLOv12 vs Grounding DINO
RTMDet vs YOLOv12 vs Grounding DINO
3 of 5Object Detection · verified 13 Aug 2026
| Specification | RTMDet | YOLOv12 | Grounding DINO |
|---|---|---|---|
| Summary | Pure throughput, and an MIT licence. | Attention-centric YOLO, still widely deployed. | Detects whatever you describe in words. |
| COCO mAP | ~52 | ~55 | ~52 zero-shot |
| Speed | 300+ | Very high | Moderate |
| Licence | MIT | AGPL-3.0 | Apache 2.0 |
| Family | Single stage | Single stage | Open vocabulary |
| Open vocabulary | No | No | Yes |
| Released | 2022 | 2025 | 2023 |
| Category | Object Detection | Object Detection | Object Detection |
| Official | OpenMMLab ↗ | Ultralytics ↗ | IDEA Research ↗ |
Highlighted rows are where these differ.
RTMDet
- Wins on raw speed where accuracy is sufficient
- MIT, so no derivative-work obligation
Best for high frame-rate video.
YOLOv12
- Attention added to the classic single-stage design
- Same licensing consideration as YOLO26
Best for teams already on the YOLO toolchain.
Grounding DINO
- Text prompt in, boxes out
- Fine-tuning still beats it on specialised objects
Best for classes you have no data for.