Requirements and Installation

This page describes the requirements and installation steps for the TLT CV Inference Pipeline.

Hardware Requirements

The TLT CV Inference Pipeline has the following hardware requirements:

Minimum

• 4 GB system RAM

• 2.5 GB of GPU RAM

• 6 core CPU

• 1 NVIDIA GPU

  • Discrete GPU: NVIDIA Volta, Turing, or Ampere GPU architecture

  • Jetson AGX Xavier, Jetson Xavier NX, or Jetson TX2

• 12 GB of HDD/SSD space

• 720p Webcam

Recommended

• 32 GB system RAM

• 32 GB of GPU RAM

• 8 core CPU

• 1 NVIDIA GPU

  • Discrete GPU: Volta, Turing, Ampere architecture

  • Jetson AGX Xavier or Jetson Xavier NX

• 16 GB of SSD space

• Webcam

  • Logitech C920 Pro HD

  • Logitech C922

  • Logitech C310

Software Requirements

The TLT CV Inference Pipeline has the following software requirements:

• Ubuntu 18.04 LTS

• NVIDIA GPU Cloud account and API key

• NVIDIA GPU Cloud CLI Tool for AMD64 or ARM64 (must exist in ${PATH})

• docker-ce with management as a non-root user using the Docker Post-Installation Steps for Linux

• nvidia docker

• NVIDIA GPU driver v455.xx or above

• Jetpack 4.5 for Jetson devices

Note

Jetpack 4.5 comes preinstalled with docker and NVIDIA Docker runtime, but will still need the Docker Post-Installation Steps for Linux.

Installation Prerequisites

Perform the following prerequisite steps before installing the TLT CV Inference Pipeline:

1. Install Docker.

2. Install NVIDIA GPU driver.

3. Install nvidia docker

4. Get an NGC account and API key. For step-by-step instructions, see the NGC Getting Started Guide.

5. Download the NVIDIA GPU Cloud CLI Tool.

6. Execute docker login nvcr.io from the command line and enter these login credentials:

   1. Username: “$oauthtoken”

   2. Password: “YOUR_NGC_API_KEY”

7. For Jetson devices, manually increase the Jetson Power mode and maximize performance further by using the Jetson Clocks mode. The following commands perform this:

   sudo nvpmodel -m 0
   sudo /usr/bin/jetson_clocks
   

Installation

The TLT CV Inference Pipeline containers and TLT models are available to download from the NGC. You must have an NGC account and an API key associated with your account. See the Installation Prerequisites section for details on creating an NGC account and obtaining an API key.

Configure the NGC API key

Using the NGC API Key obtained in Installation Prerequisites, configure the enclosed ngc cli by executing this command and following the prompts:

ngc config set

Download the TLT CV Inference Pipeline Quick Start

Now, we can download the TLT CV Inference Pipeline Quick Start using the following command:

ngc registry resource download-version "nvidia/tlt_cv_inference_pipeline_quick_start:v0.2-ga"

This will download the Quick Start Scripts, documentation for API Usage, EULA, and Third Party Licenses.

To get started, proceed to the TLT CV Inference Pipeline Quick Start Scripts.