All Projects → xiaoxiaotao → person-detection

xiaoxiaotao / person-detection

Licence: MIT License
TensorRT person tracking RFBNet300

Programming Languages

C++
36643 projects - #6 most used programming language
CMake
9771 projects
Cuda
1817 projects
c
50402 projects - #5 most used programming language

Projects that are alternatives of or similar to person-detection

deepvac
PyTorch Project Specification.
Stars: ✭ 507 (+1590%)
Mutual labels:  tensorrt, onnx
mtomo
Multiple types of NN model optimization environments. It is possible to directly access the host PC GUI and the camera to verify the operation. Intel iHD GPU (iGPU) support. NVIDIA GPU (dGPU) support.
Stars: ✭ 24 (-20%)
Mutual labels:  tensorrt, onnx
torch-model-compression
针对pytorch模型的自动化模型结构分析和修改工具集,包含自动分析模型结构的模型压缩算法库
Stars: ✭ 126 (+320%)
Mutual labels:  tensorrt, onnx
Deep Learning For Tracking And Detection
Collection of papers, datasets, code and other resources for object tracking and detection using deep learning
Stars: ✭ 1,920 (+6300%)
Mutual labels:  detection, tracking-by-detection
vs-mlrt
Efficient ML Filter Runtimes for VapourSynth (with built-in support for waifu2x, DPIR, RealESRGANv2, and Real-CUGAN)
Stars: ✭ 34 (+13.33%)
Mutual labels:  tensorrt, onnx
Paddle2onnx
PaddlePaddle to ONNX model converter
Stars: ✭ 185 (+516.67%)
Mutual labels:  detection, onnx
InsightFace-REST
InsightFace REST API for easy deployment of face recognition services with TensorRT in Docker.
Stars: ✭ 308 (+926.67%)
Mutual labels:  tensorrt, onnx
Tengine
Tengine is a lite, high performance, modular inference engine for embedded device
Stars: ✭ 4,012 (+13273.33%)
Mutual labels:  tensorrt, onnx
yolov5 tensorrt int8 tools
tensorrt int8 量化yolov5 onnx模型
Stars: ✭ 105 (+250%)
Mutual labels:  tensorrt, onnx
yolov5-deepsort-tensorrt
A c++ implementation of yolov5 and deepsort
Stars: ✭ 207 (+590%)
Mutual labels:  detection, tensorrt
Yolov5 Rt Stack
Yet another yolov5, with its runtime stack for libtorch, onnx, tvm and specialized accelerators. You like torchvision's retinanet? You like yolov5? You love yolort!
Stars: ✭ 107 (+256.67%)
Mutual labels:  detection, onnx
ONNX-Runtime-with-TensorRT-and-OpenVINO
Docker scripts for building ONNX Runtime with TensorRT and OpenVINO in manylinux environment
Stars: ✭ 15 (-50%)
Mutual labels:  tensorrt, onnx
Php Opencv Examples
Tutorial for computer vision and machine learning in PHP 7/8 by opencv (installation + examples + documentation)
Stars: ✭ 333 (+1010%)
Mutual labels:  detection, onnx
Deepstream Project
This is a highly separated deployment project based on Deepstream , including the full range of Yolo and continuously expanding deployment projects such as Ocr.
Stars: ✭ 120 (+300%)
Mutual labels:  tensorrt, onnx
YoloV5 JDE TensorRT for Track
A multi object tracking Library Based on tensorrt
Stars: ✭ 39 (+30%)
Mutual labels:  detection, tensorrt
mediapipe plus
The purpose of this project is to apply mediapipe to more AI chips.
Stars: ✭ 38 (+26.67%)
Mutual labels:  tensorrt, onnx
onnx2tensorRt
tensorRt-inference darknet2onnx pytorch2onnx mxnet2onnx python version
Stars: ✭ 14 (-53.33%)
Mutual labels:  tensorrt, onnx
Pytorch Yolov4
PyTorch ,ONNX and TensorRT implementation of YOLOv4
Stars: ✭ 3,690 (+12200%)
Mutual labels:  tensorrt, onnx
watsor
Object detection for video surveillance
Stars: ✭ 203 (+576.67%)
Mutual labels:  detection, tensorrt
YOLOX
YOLOX is a high-performance anchor-free YOLO, exceeding yolov3~v5 with MegEngine, ONNX, TensorRT, ncnn, and OpenVINO supported. Documentation: https://yolox.readthedocs.io/
Stars: ✭ 6,570 (+21800%)
Mutual labels:  tensorrt, onnx

person-detection

TensorRT person tracking RFBNet300

更新:

加入sort跟踪

原始的pytorch程序(https://github.com/songwsx/RFSong-7993.git)

大体思路:

SSD类的网络都是prior anchor和网络输出的anchor进行解码运算。为了便于运算对网络输出做了如下处理:

  1. pytorch网络输出是(7759,4)和(7759,2),为了便于运算,将其reshape成(1,7759x4+7759x2),所以网络的输出为(1,46554)
  2. 对网络输出的数据和prior anchor 一起做解码运算
  3. 图像数据输入网络,使用的是 cuda 的opencv读入gpu上,然后写个cuda程序做预处理送入网络,这么做的原因是在用cpu处理的时候速度有些慢
  4. 网络输出的数据到转换到真实的box都在cpu上运行,包括后面的nms

系统环境:

​ 1.ubuntu 18.04

​ 2.CUDA 10.2

​ 3.TensorRT 6.0

​ 4.CPU i5-8250

​ 5. GPU MX250

​ 6. opencv 3.4.6(CUDA)

运行方法:

1. 终端切换到程序根目录

2. cd cuda_lib && mkdir build && cd build && cmake .. && make
3. cd ../
4. mkdir build &&  cd build && cmake .. && make

注意事项:

  1. 注意更改main.cpp里的你要运行的视频地址
  2. 根据我的CMakeLists.txt 配置你的CMakeLists.txt
  3. cuda 的opencv 正确安装

测试:

1.测试时间: 1280 × 720视频 从模型输入到显示视频为15ms左右,此测试在MX250移动显卡上,显存消耗大概300M

TODO:

将来想法是将其推理时间减小到10ms左右,主要是模型压缩,还有将解码部分全部由gpu上运行,并且加入deep_sort进行跟踪

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].