OpenNVR 0.1.10: running an open-source NVR's AI on the Qualcomm Hexagon NPU
OpenNVR is an open-source (AGPL) NVR that records and streams from any ONVIF or RTSP camera and runs the AI model you choose behind them: YOLO detection, pose, face recognition, plate OCR, open-vocabulary detection, and a vision-language agent that can look at a scene and reason about it. It runs o
OpenNVR is an open-source (AGPL) NVR that records and streams from any ONVIF or RTSP camera and runs the AI model you choose behind them: YOLO detection, pose, face recognition, plate OCR, open-vocabulary detection, and a vision-language agent that can look at a scene and reason about it. It runs on hardware you own, air-gapped if you need, with every inference audited.
Version 0.1.10 is the first release with hardware acceleration flavours, and this post is about the one I have numbers for: Qualcomm's Hexagon NPU.
Arduino Ventuno Q: Qualcomm Dragonwing IQ-8275 (QCS8275), eight ARM cores, a Hexagon NPU, Ubuntu 24.04. Small, fanless, the kind of thing that sits behind a few cameras in a shop or a plant.
All measured on that board with the published container images. Full windows, settle times and caveats are in the benchmark doc.
In the full stack (streams, decode, tracking, the API, all on the same box):
The adapters that sit behind the detector, end-to-end on one 1080p JPEG:
The NPU's own share of an adapter call is about 9 ms. The rest is JPEG decode and preprocessing on a busy CPU, so the adapter path is now bounded by image handling, not inference.
Everything goes through ONNX Runtime's QNN execution provider, shipped as a plugin wheel ( onnxruntime-qnn ). There is no Qualcomm SDK to install on the host. The -qualcomm container gets the DSP device node, the DMA heap, and two host libraries bind-mounted, and that is it.
git clone https://github.com/open-nvr/open-nvr.git && cd open-nvr./start.sh # OPENNVR_ACCEL=auto detects /dev/fastrpc-cdsp and picks the qualcomm flavour Enter fullscreen mode Exit fullscreen mode OPENNVR_ACCEL=cpu|intel|qualcomm overrides the probe. A box with only a CPU gets exactly the images it got before.
Compiled graphs are cached, so the second start skips the ~6 s compile. The pipeline reports which provider ONNX Runtime really ran on, and a request for an accelerator that silently fell back to CPU shows up as requested!= effective in the metrics instead of being hidden.
python -m detect_pipeline.bench --latency 200 --providers CPUExecutionProvider python -m detect_pipeline.bench --latency 200 --providers QNNExecutionProvider:htp Enter fullscreen mode Exit fullscreen mode It reports median, p90 and min per device, names the provider that really ran, and exits non-zero if the accelerator fell back.
NVIDIA, desktop GPUs and Jetson. If you have a Jetson, Rockchip or Hailo box you'd lend to a benchmark, I have one board and would like more.
Expand Varun Pratap Singh Varun Pratap Singh Varun Pratap Singh Follow VoIP, WebRTC and SIP communications Expert, Extending IP Telephony to Voice AI, Scalable Voice and Agentic Voice Architectures and Deployment Strategies. Location NCR, India Education Indian Institute of Technology Kanpur Work Senior Software Engineer at Vonage Joined Sep 22, 2026 • Oct 31 Dropdown menu Copy link Hide Great OpenSource project to equip your existing IP Cameras with great power of security and AI.