Skip to content

Latest commit

 

History

History
59 lines (26 loc) · 1.84 KB

README.md

File metadata and controls

59 lines (26 loc) · 1.84 KB
Raw

Traffic Analytics using Multi Class Support in Jetson Platform Services

Notebook - Obtain spatio-temporal insights using APIs

Learning Objective:

  • Use APIs to interact with the Analytic Services on Jetson
  • Create line-crossing configuration and alerts using VST WebUI
  • Generate spatial insights like flow of vehicle movement and heatmaps

This notebook is divided into 3 parts:

  • Configure Tripwire (line-crossing) using VST WebUI

In the first section, users will experience the VST (Video Storage Toolkit) WebUI to create and configure Tripwire analytics. Not part of the lab, but users can use this WebUI to create virtual boundaries or region of interest (ROI)

  • Vehicle Movement Analytics

In this section, users will use the VST (/vst) and Analytics (/emdx) APIs to retrieve time series insights from the Analytics service. We will use this time series data to better understand the flow of vehicle movement.

  • Heatmap Analytics

In this section, user will use the behavior APIs from Analytic service to generate Heatmaps. Heatmaps provide a visualization to understand flow of traffic over time. This is generated by accumulating individual trajectories of movement and mapping it spatially over the region

  • Bring down the Workflow

In this last section, user will remove streams from VST and bring down the workflow.

Setup and Run

Configure and Run Yolov8 based App.

  • Follow the instructions in the blog 'Traffic Analytics using Multi Class Support in Jetson Platform Services' to setup the AI Perception App with Yolov8 model.

  • Add sensor in the VST UI for your camera or RTSP URL.

Install and run juypyter notebook

  • Setup a new python environment using venv

  • Activate the environment

  • pip install jupyter

  • jupyter notebook anaytics.ipynb

  • Following the instructions in the notebook