HW-SW Optimization of DNNs for Privacy-preserving People Counting on Low-resolution Infrared Arrays
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909090836381696 |
|---|---|
| author | Risso, Matteo Xie, Chen Daghero, Francesco Burrello, Alessio Mollaei, Seyedmorteza Castellano, Marco Macii, Enrico Poncino, Massimo Pagliari, Daniele Jahier |
| author_facet | Risso, Matteo Xie, Chen Daghero, Francesco Burrello, Alessio Mollaei, Seyedmorteza Castellano, Marco Macii, Enrico Poncino, Massimo Pagliari, Daniele Jahier |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_01226 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | HW-SW Optimization of DNNs for Privacy-preserving People Counting on Low-resolution Infrared Arrays Risso, Matteo Xie, Chen Daghero, Francesco Burrello, Alessio Mollaei, Seyedmorteza Castellano, Marco Macii, Enrico Poncino, Massimo Pagliari, Daniele Jahier Machine Learning Hardware Architecture 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. |
| title | HW-SW Optimization of DNNs for Privacy-preserving People Counting on Low-resolution Infrared Arrays |
| topic | Machine Learning Hardware Architecture |
| url | https://arxiv.org/abs/2402.01226 |