nndeploy is an end-to-end model inference and deployment framework. It aims to provide users with a powerful, easy-to-use, high-performance, and mainstream framework compatible model inference and deployment experience.一款端到端的模型推理和部署框架。它旨在为用户提供功能强大、简单易用、高性能且兼容主流框架的模型推理和部署体验。

an edge-real-time anchor-free object detector with decent performance

YOLOv5 in PyTorch > ONNX > RKNN

Run your yolov7 object detection with Rockchip NPU platforms (RK3566, RK3568, RK3588, RK3588S, RV1103, RV1106, RK3562).

Inference YOLOv8 segmentation on ONNX, RKNN, Horizon and TensorRT

simple yolov5 rtspserver for rk3588

A demo show how to inference blazeface model on Rockchip RV1126.

RKNN模型推理部署模板

Automated script to convert Huggingface and GGUF models to rkllm format for running on Rockchip NPU

基于u2net网络进行简单修改使其部署到rk3588板子上

export any your YOLOv7 model to TensorFlow, TensorFlowJs, ONNX, OpenVINO, RKNN,...

Inference YOLOv8 detection on ONNX, RKNN, Horizon and TensorRT

Deploy deep learning model on difference hardware and framework. (TensorRT/ONNX/MNN/RKNN)

Superresolution running on Rockchip NPU (RK3588, etc..)

Web chat front end for rk3588_npu_llm_server / RK3588 LLM chat interface

fatigue detect rknn/onnx model deploy in rk3568 npu（ROCK 3A）

瑞芯微芯片的rknn推理框架部署(yolo模型)