# NVIDIA-AI-IOT/deepstream_reference_apps

Samples for TensorRT/Deepstream for Tesla & Jetson

Repository: https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps
Canonical: https://ross.abutalabs.com/products/deepstream_reference_apps
Language: C++
License: NOASSERTION
License Family: other
Last push: 2026-07-15T09:40:36+00:00

## Health v2 (maintenance only)
Score: 74/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 92, release rhythm 35, longevity 100
- inputs: {"age_days": 2970, "days_push": 49, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1355, forks 375 (observed 2026-08-28T04:04:29.157529+00:00)

## What it is
A collection of C++ reference applications demonstrating video analytics pipelines built on NVIDIA DeepStream SDK 9.0 and TensorRT for Tesla GPUs and Jetson edge devices. The samples cover anomaly detection, runtime source add/delete, SAM2-based multi-object tracking and segmentation, single- and multi-view 3D tracking, parallel multi-model inference, 3D body pose inference, and IPC video buffer sharing. The repository has ceased updates in favor of the main NVIDIA/deepstream repository.

## Use cases
- deepstream sample apps for jetson video analytics
- multi-object tracking and segmentation with sam2 on deepstream
- single-view or multi-view 3d tracking across calibrated cameras
- run multiple models in parallel inference with deepstream
- add and remove video sources at runtime in a deepstream pipeline
- video anomaly detection reference implementation
- share video buffers between gstreamer pipelines over ipc
- 3d body pose estimation inference with tensorrt

## When to choose
- You are building video analytics on NVIDIA Jetson or Tesla GPUs with DeepStream 9.0 and want working C++ starting points
- You need reference implementations for multi-object tracking, SAM2 segmentation, 3D tracking, or bodypose model inference
- You want examples of advanced pipeline patterns like dynamic source management, parallel multi-model inference, and IPC buffer sharing

## When to avoid
- You need the latest DeepStream features or ongoing updates - this repo has ceased updates and NVIDIA/deepstream is the successor
- You are not using NVIDIA GPUs, Jetson hardware, or the DeepStream SDK
- You need a production-ready turnkey product rather than sample and reference code

## Facets
- artifact type: learning-resource
- maturity: maintenance
- function: video-processing, computer-vision, deep-learning, machine-learning, gpu-computing
- domain: computer-vision, artificial-intelligence, deep-learning, embedded-systems, tutorials
- platform: cpp, embedded
- tags: deepstream, tensorrt, jetson, nvidia, video-analytics, sample-applications, multi-object-tracking, sam2-segmentation, 3d-tracking, pose-estimation, anomaly-detection, gstreamer, video, linux, gpu

## Member repositories
- NVIDIA-AI-IOT/deepstream_reference_apps (main) score 74

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:29.157529+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:41:56.866278+00:00, confidence not recorded.
  - readme: https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps (fetched 2026-08-28T04:04:29.157529+00:00, sha fc1d7e25f756)
- Data as of 2026-08-30T08:39:29.467469+00:00.