Andes Technology and Deeplite Inc. are excited to announce the latest results in their partnership for AI-powered applications using Deeplite’s unique optimization software and Andes’ low-power Andes RISC-V CPU cores.
“We are determined to provide the most efficient and accurate solutions possible for low-power devices, particularly as edge AI is increasingly deployed in smart assistants, security cameras and smart manufacturing applications,” said Dr. Charlie Su, CTO and Executive VP of Andes Technology.
This partnership focuses on compressing and accelerating the Visual Wake Words (VWW) application, where a tiny embedded camera performs person detection. Together, Deeplite and Andes produced various optimized INT8 models from the floating-point based Mobilenet-v1-0.25x model and achieved groundbreaking, industry-leading results.
As an outcome, we delivered two sets of results. The first set concentrated on increasing the accuracy, where we achieved 2.7% higher accuracy, 1.7x (172 KB) smaller size, and 9% faster execution. For the second set of results, Deeplite and Andes focused on maximizing the compression and achieved 2.3x (121KB) smaller size, slightly higher accuracy (0.7%), and 15% faster execution when compared to the INT8 model provided by TensorFlow Lite Micro.