Intel Movidius Neural Compute Stick | New Product BriefOctober 30, 2017 by Mouser Electronics
Intel’s Movidius Neural Compute Stick is a USB stick for development and offline deployment of deep learning applications to edge devices.
This New Product Brief (NPB) is part of a video series highlighting the features, applications, and technical specs of newly-released products.
Intel Movidius Neural Compute Stick
Intel’s Movidius Neural Compute Stick is a USB stick for development and offline deployment of deep learning applications to edge devices. The Movidius NCS is based on the Myriad 2 Visual Processing Unit, providing over 100 gigaflops within a 1 W power envelope and enables real-time, on-device inference. Developers can use the SDK to compile trained Caffe networks to run on the NCS VPU, then tune and validate to optimize for real-world performance. This reduces latency, lowers power consumption, and enhances privacy in machine intelligence for smart home, security, retail, and healthcare applications. The Movidius Neural Compute Stick can also act as a scalable, self-contained neural network accelerator by adding multiple devices to a platform, bringing improved performance and power efficiency.
- Supports Convolutional Neural Network (CNN) profiling, prototyping and tuning workflow
- Quickly deploy existing CNN models or uniquely trained networks
- Real-time, on-device inference - cloud connectivity not required
- Run multiple devices on the same platform to scale performance
- Movidius VPU with energy-efficient CNN processing
- Dimensions: 72.5x27x14mm
New Industry Products are a form of content that allows industry partners to share useful news, messages, and technology with All About Circuits readers in a way editorial content is not well suited to. All New Industry Products are subject to strict editorial guidelines with the intention of offering readers useful news, technical expertise, or stories. The viewpoints and opinions expressed in New Industry Products are those of the partner and not necessarily those of All About Circuits or its writers.