/jetson_trt_yolo

research project while in University of Stuttgart

Primary LanguageC++

Object detection to prevent finger injuries

Object detection applications to prevent finger injuries in a cutting tool holder environment

research project while in University of Stuttgart

avatar

Features:

Customized detection area

LED lights up when a finger is detected

Detection speed of about 40fps

Background

Inspired by the following video. (Youtube) The idea is in protecting the safety of the chainsaw operator. When the saw is working, customize the danger zone. If a finger is found in the dangerzone, a signal is output to alert the operator. SawStop | SawStop Hot Dog Demo

Hardware

  1. Jetson Nano Developer Kit 4GB
  2. Micro SD card above 32GB
  3. USB Webcam
  4. Breadboard, jumper wires, resistance and LED (for GPIO)

Setup

follow the official guide to boot Jetson Nano Getting Started with Jetson Nano Developer Kit

Library Installation

1. update and upgrade

sudo apt-get update
sudo apt-get upgrade

2. add cuda PATH

cd ~
gedit .bashrc

Add these three lines at the end, Ctrl+S to save and close gedit

export PATH=/usr/local/cuda/bin:$PATH

export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

export CUDA_ROOT=/usr/local/cuda

Activate and reboot

source .bashrc
reboot

3. install pytorch 1.8 and torchvision v0.9.0 via PyTorch for Jetson - version 1.10 now available

wget https://nvidia.box.com/shared/static/p57jwntv436lfrd78inwl7iml6p13fzh.whl -O torch-1.8.0-cp36-cp36m-linux_aarch64.whl
sudo apt-get install python3-pip libopenblas-base libopenmpi-dev 
pip3 install Cython
pip3 install numpy torch-1.8.0-cp36-cp36m-linux_aarch64.whl

sudo apt-get install libjpeg-dev zlib1g-dev libpython3-dev libavcodec-dev libavformat-dev libswscale-dev
git clone --branch v0.9.0 https://github.com/pytorch/vision torchvision   
cd torchvision
export BUILD_VERSION=0.9.0 
python3 setup.py install --user

4. install yolov5

install dependency

sudo apt-get install liblapack-dev
sudo apt-get install libblas-dev
sudo apt-get install gfortran

install scipy matplotlib pillow pyyaml tensorboard tqdm

pip3 install scipy matplotlib pillow pyyaml tensorboard tqdm

install pillow correctly

pip3 uninstall pillow
pip3 install pillow --no-cache-dir
  1. install pycuda library
export PATH=/usr/local/cuda-10.2/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-10.2/lib64:$LD_LIBRARY_PATH
pip3 install pycuda --user

Inference

cd ~
git clone https://github.com/newbiehyz/hand_jetsonnano
cd hand_jetsonnano
python3 detect.py   # press 'q' to exit