The ZIA DV700 Series is a highly precise AI processing component tailored for advanced systems that demand top-tier reliability such as autonomous vehicles and robotics. This series incorporates standard support for FP16 precision floating-point operations which enables seamless integration without the need for model retraining, perfect for AI models developed on PC or cloud servers. This feature facilitates the system's ability to maintain high inference accuracy crucial in critical AI applications.
Designed with deep learning in mind, the ZIA DV700 Series provides an optimal hardware architecture for inference tasks. Its hard-wired structure makes it ideal for detection tasks, semantic segmentation, skeleton estimation, and other varied DNN models by leveraging the power of Mobilenet, Yolo v3, SegNet, and PoseNet.
Development is streamlined with the inclusion of a support environment (SDK/Tool), enabling smooth integration with AI frameworks such as Caffe, Keras, and TensorFlow. Once models are prepared under these frameworks, developers can effortlessly migrate AI inference tasks to the DV700 Series.
