dfine

Accelerating SAHI-based inference on YOLO models using TensorRT.

C++ app for computer vision inference, supporting multiple tasks and backends.

C++ application to perform computer vision tasks using Nvidia Triton Server for model inference

Unified PyTorch implementations of RT-DETRv3, D-FINE, DEIM, RT-DETRv4 & DEIMv2 — 30 COCO variants, one CLI for train/eval/infer/export. DETR 系列实时目标检测模型统一 PyTorch 实现

TensorRT-DETR is a C++/CUDA/TensorRT inference deployment library for NVIDIA GPUs. It provides C++ and Python APIs for detection, instance segmentation, pose estimation.

Master's project comparing SOTA models for object detection on a custom-built geospatial dataset. Implements YOLOv12, DFine, and RF-DETR.