Skip to content

README.md

Latest commit

 

History

History
79 lines (57 loc) · 2.42 KB

README.md

File metadata and controls

79 lines (57 loc) · 2.42 KB

Appendix: AI NVR on reComputer

Why This Appendix Exists

The main 10-section course is designed around a clean learning spine:

  • foundations
  • methods
  • tasks
  • training
  • deployment
  • Jetson system concepts
  • frontier outlook

An AI NVR project is extremely valuable, but it works best as an extension after the learner has already completed the main sequence. That is why it appears here as an appendix rather than as one of the main 10 sections.

Project Goal

Build an AI NVR workflow on reComputer that combines:

  • camera ingestion
  • live inference
  • tracking
  • event logic
  • recording and storage
  • operator-facing visualization

What This Project Brings Together

This appendix is useful because it combines ideas from multiple earlier chapters:

  • 4.2: image and video input
  • 4.5: vision tasks such as detection and tracking
  • 4.6: model usage and evaluation
  • 4.7: deployment formats and optimization
  • 4.8: pipeline thinking
  • 4.9: DeepStream and Jetson services

Recommended Architecture

  • VST for stream onboarding and video handling
  • DeepStream Perception for inference
  • Analytics for ROI, counting, or line-crossing logic
  • Redis for metadata
  • Ingress for service access

Minimal Project Sequence

  1. prepare the Jetson baseline
  2. verify JPS services
  3. add one or more camera streams
  4. launch inference
  5. confirm tracking or event logic
  6. inspect recordings and storage behavior
  7. validate the operator experience

Example Service Startup

sudo systemctl start jetson-redis
sudo systemctl start jetson-ingress
sudo systemctl start jetson-vst

Reflection Questions

  1. Which parts of the project depend on model quality, and which depend on system design?
  2. What would fail first in a real deployment: the model, the stream, the storage, or the operator workflow?
  3. How would you extend this appendix into a production pilot?

Summary

This appendix gives learners a concrete project that connects the course's theory and system practice. It is not the center of the curriculum, but it is a powerful extension for learners who want to see how a complete edge vision application comes together.

References