Artificial intelligence optical hardware empowers high-resolution hyperspectral video understanding at 1.2 Tb/s

TL;DR

This paper presents an innovative optoelectronic hardware platform that enables real-time hyperspectral video understanding at 1.2 Tb/s, significantly surpassing existing technologies in speed and spectral resolution.

Contribution

The work introduces a novel hardware-accelerated integrated platform combining optical AI processing with machine vision, achieving unprecedented data throughput for multidimensional video analysis.

Findings

Achieves 1.2 Tb/s processing speed for hyperspectral video

Surpasses similar technologies by three to four orders of magnitude

Validates performance in semantic segmentation and object understanding tasks

Abstract

Foundation models, exemplified by GPT technology, are discovering new horizons in artificial intelligence by executing tasks beyond their designers' expectations. While the present generation provides fundamental advances in understanding language and images, the next frontier is video comprehension. Progress in this area must overcome the 1 Tb/s data rate demanded to grasp real-time multidimensional video information. This speed limit lies well beyond the capabilities of the existing generation of hardware, imposing a roadblock to further advances. This work introduces a hardware-accelerated integrated optoelectronic platform for multidimensional video understanding in real-time. The technology platform combines artificial intelligence hardware, processing information optically, with state-of-the-art machine vision networks, resulting in a data processing speed of 1.2 Tb/s with hundreds of frequency bands and megapixel spatial resolution at video rates. Such performance, validated in the AI tasks of video semantic segmentation and object understanding in indoor and aerial applications, surpasses the speed of the closest technologies with similar spectral resolution by three to four orders of magnitude. This platform opens up new avenues for research in real-time AI video understanding of multidimensional visual information, helping the empowerment of future human-machine interactions and cognitive processing developments.