Transforming Far-Edge Computer Vision With Energy-Efficient AI
tinyRAPTOR's AI stack for embedded vision
![](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/chip-2.png?lossy=1&strip=1&webp=1)
tinyRAPTOR is a digitally scalable silicon proven NPU* hardware IP.
Its convenient SDK accelerates Deep Neural Network exploration and minimizes inference implementation costs.
Our pre-optimized AI models for memory-constrained chips deliver accurate Computer Vision insights at more than 20 frames per second yet under 1 mW processing power.
*NPU: Neural Processing Unit
see tinyRAPTOR's solution in action on our device-like demonstrator
![](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/Female-virtual-reality-1.jpg?lossy=1&strip=1&webp=1)
Want to know our secret sauce to deliver first class computation-rate at every milli-watt ?
tinyRAPTOR’s Near Memory Computing NPU architecture and its inference-efficient AI stack optimize data flow processing by eliminating energy-wasting data movement.
Always ON Computer Vision is now possible.
Enable countless new use-cases with Always-ON Computer Vision
![Classification & Detection Classification & Detection](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/Classification-Detection-2.jpg?lossy=1&strip=1&webp=1)
Classification & Detection
![Segmentation Segmentation](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/Segmentation-3.jpg?lossy=1&strip=1&webp=1)
Segmentation
identify multiple objects, understand their shapes & boundaries.
Empower your clients
to deploy classification, detection or segmentation AI models on tinyRAPTOR’s enabled SoCs or Image Sensors, to deliver carbon-friendly vision intelligence for endless use-cases.
Push button AI deployment
turn data science into production ready software in a few days
![](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/tinyRaptorSDK-1.jpg?lossy=1&strip=1&webp=1)
The tinyRAPTOR's SDK enables rapid exploration & implementation of Deep Neural Networks into tinyRAPTOR's enabled SoC or Image Sensors.
![](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/laptop.png?lossy=1&strip=1&webp=1)
Start from the comfort of your preferred development environment such as Data Science Notebooks or Visual Studio Code,
Sweep through a range of tinyRAPTOR’s frequencies, voltages & MAC configurations,
And find-out the ideal energy-latency sweet-spot for your AI model optimized for tinyRAPTOR’s Neural Processing Unit.
Pair tinyRAPTOR’s SDK together with NAS* tools, spot pareto-front families of AI models, choose one suiting your use-case
*NAS: Network Architecture Search
![](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/chart-2.png?lossy=1&strip=1&webp=1)
Try tinyRAPTOR's SDK now,
and push AI compute capabilities right at the end-point.
pre-optimized AI models
![](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/github-Ai-model-zoo.png?lossy=1&strip=1&webp=1)
tinyRAPTOR’s Vision AI Models Zoo is a collection of pre-optimized AI models for computer visions tasks such as Classification, Detection, Segmentation
![](https://b3338709.smushcdn.com/3338709/wp-content/uploads/2023/08/tinyRAPTORK-circle.png?lossy=1&strip=1&webp=1)
Use our optimized AI models for tinyRAPTOR’s NPU hardware, as a launchpad to create customized models fine-tuned to your final application use-case.