Turning old models fashion again: Recycling classical CNN networks using the Lattice Transformation

TL;DR

This paper proposes a novel method called LCNN cross-fusion to recycle and enhance classical CNN architectures, significantly improving their accuracy on image classification tasks by leveraging lattice transformations.

Contribution

It introduces the LCNN cross-fusion strategy to adapt and improve old CNN models, demonstrating substantial accuracy gains on benchmark datasets.

Findings

Achieved up to 63.21% accuracy increase on NORB dataset

Reused classical CNN architectures with minimal modifications

Identified limitations and disadvantages of the proposed strategy

Abstract

In the early 1990s, the first signs of life of the CNN era were given: LeCun et al. proposed a CNN model trained by the backpropagation algorithm to classify low-resolution images of handwritten digits. Undoubtedly, it was a breakthrough in the field of computer vision. But with the rise of other classification methods, it fell out fashion. That was until 2012, when Krizhevsky et al. revived the interest in CNNs by exhibiting considerably higher image classification accuracy on the ImageNet challenge. Since then, the complexity of the architectures are exponentially increasing and many structures are rapidly becoming obsolete. Using multistream networks as a base and the feature infusion precept, we explore the proposed LCNN cross-fusion strategy to use the backbones of former state-of-the-art networks on image classification in order to discover if the technique is able to put these designs back in the game. In this paper, we showed that we can obtain an increase of accuracy up to 63.21% on the NORB dataset we comparing with the original structure. However, no technique is definitive. While our goal is to try to reuse previous state-of-the-art architectures with few modifications, we also expose the disadvantages of our explored strategy.