The UP Squared* AI Vision Developer Kit comes preconfigured to provide heterogeneous processing across Intel® hardware and accelerators. The development kit enables developers to create computer vision (CV) solutions with preinstalled software packages including:
- Intel® Distribution of OpenVINO™ toolkit, a convolutional neural network (CNN) toolkit
- Intel® System Studio, a cross-platform tool suite for improving performance on Intel® platforms
- Arduino Create*, an IDE for creating cloud-based C++ solutions
In addition, the software offerings include a pair of deep learning inference tutorials, available for download. These tutorials enable developers to experiment with a sample application and pretrained detection models. The tutorial packages contain sample video to demonstrate functionality. Both tutorials are available for command line, Intel System Studio, and Arduino Create on GitHub*.
The tutorials involve running inference on pretrained models on CPU, GPU, and VPU devices. The sample application of each tutorial accepts video from USB cameras, recorded video files, or still image files and sends output to an image window.
Face detection tutorial
The face detection tutorial uses a pretrained convolutional neural network (CNN) with a sample application and sample video to demonstrate facial recognition.
The application detects human faces, age, and gender.
The car detection tutorial uses a pretrained convolutional neural network (CNN) with a sample application and sample video to demonstrate recognition of vehicles.
The application detects vehicle type and color.
The Intel® Distribution of OpenVINO™ toolkit contains:
- A Model Optimizer that optimizes pretrained models and converts them to an intermediate representation (IR)
- An Inference Engine that accepts optimized models and permits testing in target environments
- An Intel® Deep Learning Deployment Toolkit which encompasses the Model Optimizer and Inference Engine and deploys to target environments
The development and training of models is beyond the scope of the tutorials. For details about using your trained models with the Inference Engine, see Model Optimizer Developer Guide and Inference Engine Developer Guide.
See additional deep learning resources to understand more about the software architectural components in the UP Squared* AI Vision Developer Kit.