The software provides 3D object detection and tracking using PAL USB. PAL- USB is the only single sensor omnidirectional vision system to provide 360° stereoscopic sensing with depth perception.

Please follow the instructions given below on any of the Nvidia Jetson embedded boards with Jetpack 4.6 to install the software.

 sudo apt-get install git-lfs
 git clone -b PAL --single-branch https://github.com/DreamVu/3D-Object-Detection-Tracking.git
 cd 3D-Object-Detection-Tracking
 git lfs pull

Confirm the following dependencies. These are must have to proceed further

• Jetpack 4.6

• CUDA enabled OpenCV 4.4.0 and OpenCV Contrib 4.4.0 libraries.

  Follow these steps to install the required OpenCV dependencies.

•  cd installations
   chmod +x ./*.sh
   ./opencv.sh
  

• Python 3.6 libraries (pytorch, torchvision, numpy, PIL, etc.)

  cd installations
  chmod +x ./*.sh
  ./install.sh arg1 arg2

• arg1: It can be either 0, 1 or can be skipped. On selecting 1, installation will build the whole software based on the particular Nvidia Jetson architecture and will enable higher performance. This may take a few hours to complete the installation. On selecting 0, it will quickly build the software by using some of the pre-configured libraries provided. If arg1 is skipped, then arg2 must also be skipped.

• arg2: If the arg1 is 1, then the second argument takes in the amount of RAM the process will use during the build of the software in MBs. Following are the recommended values for different Nvidia Jetson architectures. For Jetson Xavier NX arg2: 3500 For Jetson Xavier AGX arg2: 8000 For Jetson Nano arg2: 1000

Once complete please reboot the system.

For rest of the evaluation of the 3D Object Detection and Tracking SDK, please read the

• Code Samples

If you need help or more informations check our Help Center or join our Community.

Feel free to open an issue if you find a bug, or a pull request for bug fixes, features or other improvements.