agentleFS
Sign inStart free
Seeed-ProjectsSeeed-Projects/Seeed-Jetson-DevelopTool/skills/claude/traffic-deepstream/CLAUDE.md

Seeed-Jetson-DevelopTool / traffic-deepstream

CLAUDE.md57 starsChanged 4 months agoMITView on GitHub

Deploy an intelligent traffic management system using NVIDIA DeepStream SDK on Jetson devices, with DashCamNet for vehicle detection and VehicleTypeNet/VehicleMakeNet for classification. Tested on reComputer J1010 with JetPack 4.6.1.

  • Installs packages

The file

We split this CLAUDE.md into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run one phase at a time. After each phase, verify the expected result before continuing.

    line 10
  2. Very specific:

    If a phase succeeds → print [OK] and move to the next phase.

    line 11
  3. Very specific:

    If a phase fails → print [STOP], consult the failure decision tree, and ask the user before retrying.

    line 12

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

Traffic Management with DeepStream SDK

Execution model

Run one phase at a time. After each phase, verify the expected result before continuing.

  • If a phase succeeds → print [OK] and move to the next phase.
  • If a phase fails → print [STOP], consult the failure decision tree, and ask the user before retrying.

Phase 1 — Verify prerequisites

Required:

  • Jetson device with JetPack installed (tested on reComputer J1010 + JetPack 4.6.1)
  • SDK Components and DeepStream SDK installed
  • Keyboard, HDMI display
  • USB webcam or MIPI CSI camera
cat /etc/nv_tegra_release
dpkg -l | grep deepstream

Expected: L4T version shown; DeepStream SDK package listed.

Phase 2 — Download configuration files

git clone https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps.git
cd deepstream_reference_apps/deepstream_app_tao_configs/
sudo cp -a * /opt/nvidia/deepstream/deepstream/samples/configs/tao_pretrained_models/

Expected: Config files copied to DeepStream samples directory.

Phase 3 — Download pre-trained models

sudo apt install -y wget zip
cd /opt/nvidia/deepstream/deepstream/samples/configs/tao_pretrained_models/
sudo ./download_models.sh

Expected: DashCamNet, VehicleMakeNet, and VehicleTypeNet models downloaded.

Phase 4 — Configure DeepStream application

Edit the configuration file:

cd /opt/nvidia/deepstream/deepstream/samples/configs/tao_pretrained_models/
sudo vi deepstream_app_source1_dashcamnet_vehiclemakenet_vehicletypenet.txt

Make these changes:

  1. Under [sink0], change sync=1 to sync=0
  2. Under [primary-gie], set model-engine-file=../../models/tao_pretrained_models/dashcamnet/resnet18_dashcamnet_pruned.etlt_b1_gpu0_fp16.engine
  3. Under [secondary-gie0], set model-engine-file=../../models/tao_pretrained_models/vehiclemakenet/resnet18_vehiclemakenet_pruned.etlt_b4_gpu0_fp16.engine
  4. Under [secondary-gie1], set model-engine-file=../../models/tao_pretrained_models/vehicletypenet/resnet18_vehicletypenet_pruned.etlt_b4_gpu0_fp16.engine

Phase 5 — Run the traffic management demo

Connect camera, keyboard, and HDMI display to the Jetson device.

cd /opt/nvidia/deepstream/deepstream/samples/configs/tao_pretrained_models/
sudo deepstream-app -c deepstream_app_source1_dashcamnet_vehiclemakenet_vehicletypenet.txt

Expected: Live video with vehicle detection, type classification, and make classification displayed on HDMI.

To try other demos:

sudo deepstream-app -c deepstream_app_source1_$MODEL.txt

Failure decision tree

SymptomLikely causeSuggested fix
deepstream-app not foundDeepStream SDK not installedInstall via NVIDIA SDK Manager
Model download failsNo internet or script errorCheck connectivity; run download_models.sh manually
Engine file not foundFirst run needs TensorRT engine buildWait for engine generation on first launch (may take several minutes)
No display outputWrong sink type or sync settingVerify type=2 (EglSink) and sync=0 in [sink0]
Camera not detectedUnsupported camera or wrong source configCheck v4l2-ctl --list-devices; update [source0] in config
Low FPSGPU not fully utilizedEnsure batch-size is appropriate; check tegrastats for GPU usage

Reference files

  • references/source.body.md — Full original wiki content (reference only)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free