Vertigo – AI for IoT & The Edge
AI gets interesting, when the increasing efficiency of the AI algorithms meets the increasing power of IoT SoCs and sensors that can be deployed at the edge, by anyone, everywhere at negligible cost. Utilizing our proprietary breakthrough-technology, our deep learning vision networks offer state-of-the-art accuracy, while being able to run at high speed on commodity off-the-shelf sensors smaller than a Raspberry Pi, or at up to 30fps when ingesting 4k video on a $59 NVIDIA Jetson Nano, or over 100fps on a modern iPhone or IPad.
Our custom sparsity-optimized Vertigo AI Inference Engine offers:
- A self-contained, complete AI pipeline with Python, Swift, Go and C++ APIs
- Precise head-counting for demanding indoor and outdoor applications.
- 99.5 unconstrained face recognition accuracy on the LFW benchmark at high frame rates
- Cross-platform support for all major processors and GPUs
- OpenMP, Apple Dispatch/Metal, Windows D3D, OpenCL, CUDA, AVX2/512, NEON acceleration support.
- A GDPR compliant privacy-preserving solution that runs independently at the edge.
- Offline and volume licensing available.
Our most recent innovation is the Focus video codec and conferencing platform.
Below are some older demos of libvertigo in use, please contact us at firstname.lastname@example.org for more demos or licensing information.