Low-resolution infrared (IR) array sensors enable people counting applications such as monitoring the occupancy of spaces and people flows while preserving privacy and minimizing energy consumption. Deep Neural Networks (DNNs) have been shown to be well-suited to process these sensor data in an accurate and efficient manner. Nevertheless, the space of DNNs' architectures is huge and its manual exploration is burdensome and often leads to sub-optimal solutions. To overcome this problem, in this work, we propose a highly automated full-stack optimization flow for DNNs that goes from neural architecture search, mixed-precision quantization, and post-processing, down to the realization of a new smart sensor prototype, including a Microcontroller with a customized instruction set. Integrating these cross-layer optimizations, we obtain a large set of Pareto-optimal solutions in the 3D-space of energy, memory, and accuracy. Deploying such solutions on our hardware platform, we improve the state-of-the-art achieving up to 4.2x model size reduction, 23.8x code size reduction, and 15.38x energy reduction at iso-accuracy.

翻译：低分辨率红外阵列传感器支持人数统计应用，例如监测空间占用率和人员流动，同时保护隐私并最大限度地降低能耗。深度神经网络已被证明能够准确高效地处理这些传感器数据。然而，深度神经网络的架构空间极为庞大，人工遍历不仅繁琐，且往往导致次优解。为解决这一问题，本文提出了一种高度自动化的深度神经网络全栈优化流程，涵盖神经架构搜索、混合精度量化、后处理，直至实现包含定制指令集微控制器的新型智能传感器原型。通过集成这些跨层优化，我们在能量、内存和精度构成的三维空间中获得了大量帕累托最优解。将这些解部署到我们的硬件平台上，我们在等精度条件下实现了高达4.2倍的模型大小缩减、23.8倍的代码大小缩减以及15.38倍的能量消耗缩减，从而改进了现有技术水平。