Recurrent Neural Networks for Still Images

TL;DR

This paper demonstrates that Recurrent Neural Networks can be effectively adapted for still image processing, offering a resource-efficient alternative to CNNs and transformers, especially for embedded systems.

Contribution

The paper introduces a novel RNN design for 2D images and a memory-efficient BiRNN variant, enhancing image analysis in compact models.

Findings

RNN-based models outperform traditional methods on COCO and CIFAR100 datasets.

Proposed RNN architectures are more memory-efficient for embedded systems.

RNNs can effectively interpret image pixels as sequences for improved performance.

Abstract

In this paper, we explore the application of Recurrent Neural Network (RNN) for still images. Typically, Convolutional Neural Networks (CNNs) are the prevalent method applied for this type of data, and more recently, transformers have gained popularity, although they often require large models. Unlike these methods, RNNs are generally associated with processing sequences over time rather than single images. We argue that RNNs can effectively handle still images by interpreting the pixels as a sequence. This approach could be particularly advantageous for compact models designed for embedded systems, where resources are limited. Additionally, we introduce a novel RNN design tailored for two-dimensional inputs, such as images, and a custom version of BiDirectional RNN (BiRNN) that is more memory-efficient than traditional implementations. In our research, we have tested these layers in Convolutional Recurrent Neural Networks (CRNNs), predominantly composed of Conv2D layers, with RNN layers at or close to the end. Experiments on the COCO and CIFAR100 datasets show better results, particularly for small networks.