- EOL
|
Attention! The product has been discontinued. Check other products in category. |
Description: Intel Neural Compute Stick 2 - USB neural network
The small USB module, which looks like a flash drive, allows you to create prototypes of Internet of Things devices, teach AI and deploy DNN deep neural network applications. The device workswithout a connection to the cloud.It supports heterogeneous performance in computer vision accelerators:CPU, GPU, VPU and FPGAusing common API. The device is used in facial, object and even emotion recognition projects. Intel NeuralCompute Stickworks with Windows, macOS and Linux devices, including Raspberry Pi mini-computer.
Installation in Raspbian system.
New features of Intel Neural Compute Stick 2
- Built-in deep learning tool.
- Supports multi-device inference for load management.
- Offers methods of binary distribution through package managers and Docker containers.
- Provides new APIsoriented to the inference engine.
USB Intel Compute Stick 2 neural network module.
Intel Movidus Myriad X Vision processing unit
The latest generation Intel VPU has 16 high-performance SHAVE cores and a dedicated deep network hardware accelerator for superior viewing quality and an AI application interface.
Intel Compute Stick 2 module.
Intel Neural Compute Stick 2 specification
- CPU: Intel Movidius Myriad X VPU
- Supported frameworks: TensorFlow, Caffe, Apache MXNet, Open Network Exchange (ONXX), PayTorch, PaddlePaddle by ONNX conversions
- Communication: USB 3.0 type A
- Toolkit: OpenVINO
- Supported operating systems:
- Ubuntu 16.04.3 LTS (64-bit)
- CentOS 7.4 (64-bit)
- Windows 10 (64-bit)
- macOS 10.14.4 (or newer)
- Raspbian
- Others (through OpenVINO open source distributions)
- Operating temperature: 0°C to 40°C
- Dimensions: 72.5 x 27 x 14 mm
Explore the capabilities of Intel Neural Compute Stick 2
Intel Neural Compute Stick 2 - USB neural network is a small Intel-branded computing module that acts as a neural coprocessor for Windows, macOS or Linux computers. The module is compatible with both classic computers and Raspberry Pi series minicomputers.
The module does not require a connection to the cloud to work. It offers support for well-known frameworks and libraries for creating machine learning applications - including TensorFlow, Apache MXNet, Open Network Exchange (ONXX) or PyTorch.
Intel Neural Compute Stick 2 coprocessor applications
The module is designed to support the host computer while performing calculations on an artificial neural network. This type of deep learning algorithms is currently one of the most common applications of artificial intelligence, mainly in applications related to image processing - face recognition, posture tracking or analysis of the environment around the autonomous vehicle.
However, the scope of applications is not limited to vision systems. Also neural network supported sensors or sensor arrays, generating large amounts of data in real time, can be successfully used. Neural networks are perfectly suited for the classification of learned patterns as well as for the search for new information - hence their increasing use in data minig techniques.
Useful links |
