Human-Detection-using-Jetson-Inference

Jetson inference is realtime DNN vision library for NVIDIA Jetson Nano/TX1/TX2/Xavier NX/AGX Xavier.

For human detection, we can use some models such as pednet, multiped, SSD MobileNet-V1, SSD MobileNet-V2, and SSD Inception-V2.

Installation

Install some requirements

$ sudo apt-get install libboost-all-dev
$ sudo apt-get install git cmake libpython3-dev python3-numpy
$ sudo apt-get install build-essential python-dev python-setuptools libboost-python-dev libboost-thread-dev

Install Jetson Inference

$ sudo apt-get update
$ git clone --recursive https://github.com/dusty-nv/jetson-inference
$ cd jetson-inference
$ mkdir build
$ cd build
$ cmake ../
$ make -j$(nproc)
$ sudo make install
$ sudo ldconfig

Select model:

• Ped-100
• Multi-500
• SSD-MobileNet-V1
• SSD-MobileNet-V2
• SSD-Inception-V2

More about Jetson Inference: https://github.com/dusty-nv/jetson-inference

Human Detection Comparison Results

Real-Time Human Detection Using Deep Learning on Embedded Platforms: A Review

https://journal.umy.ac.id/index.php/jrc/article/view/10558