ImageSig: A signature transform for ultra-lightweight image recognition

TL;DR

ImageSig introduces a lightweight, signature-transform-based method for image recognition that achieves high accuracy with minimal computational resources, suitable for embedded devices and edge AI applications.

Contribution

The paper presents a novel image recognition approach using signature transforms that requires no convolutional or attention mechanisms, enabling ultra-lightweight models with high accuracy.

Findings

Outperforms many state-of-the-art methods on 64x64 RGB images

Requires significantly less FLOPS, power, and memory

Pretrained models as small as 44.2 KB are effective on embedded hardware

Abstract

This paper introduces a new lightweight method for image recognition. ImageSig is based on computing signatures and does not require a convolutional structure or an attention-based encoder. It is striking to the authors that it achieves: a) an accuracy for 64 X 64 RGB images that exceeds many of the state-of-the-art methods and simultaneously b) requires orders of magnitude less FLOPS, power and memory footprint. The pretrained model can be as small as 44.2 KB in size. ImageSig shows unprecedented performance on hardware such as Raspberry Pi and Jetson-nano. ImageSig treats images as streams with multiple channels. These streams are parameterized by spatial directions. We contribute to the functionality of signature and rough path theory to stream-like data and vision tasks on static images beyond temporal streams. With very few parameters and small size models, the key advantage is that one could have many of these "detectors" assembled on the same chip; moreover, the feature acquisition can be performed once and shared between different models of different tasks - further accelerating the process. This contributes to energy efficiency and the advancements of embedded AI at the edge.