D-FINE

D-FINE is a real-time object detection model introduced in the paper " D-FINE: Redefine Regression Task of DETRs as Fine‑grained Distribution Refinement".

D-FINE introduces two architecture features:

1. Fine-grained distribution refinement (FDR), and;
2. Global optimal localization self-distillation (GO-LSD).

In the paper abstract, the authors note the following results:

Specifically, D-FINE-L / X achieves 54.0% / 55.8% AP on the COCO dataset at 124 / 78 FPS on an NVIDIA T4 GPU. When pretrained on Objects365, D-FINE-L / X attains 57.1% / 59.3% AP, surpassing all existing real-time detectors. Furthermore, our method significantly enhances the performance of a wide range of DETR models by up to 5.3% AP with negligible extra parameters and training costs.