Progressive Meta-Pooling Learning for Lightweight Image Classification Model

TL;DR

This paper introduces Meta-Pooling, a learnable receptive field method for lightweight networks, improving accuracy on ImageNet by adaptively enlarging receptive fields without increasing model complexity.

Contribution

It proposes a novel Meta-Pooling framework with a Progressive learning strategy to optimize receptive fields in lightweight neural networks.

Findings

MobileNetV2 with Meta-Pooling achieves 74.6% top-1 accuracy on ImageNet.

Meta-Pooling outperforms baseline MobileNetV2 by 2.3%.

The method effectively learns suitable receptive fields for improved performance.

Abstract

Practical networks for edge devices adopt shallow depth and small convolutional kernels to save memory and computational cost, which leads to a restricted receptive field. Conventional efficient learning methods focus on lightweight convolution designs, ignoring the role of the receptive field in neural network design. In this paper, we propose the Meta-Pooling framework to make the receptive field learnable for a lightweight network, which consists of parameterized pooling-based operations. Specifically, we introduce a parameterized spatial enhancer, which is composed of pooling operations to provide versatile receptive fields for each layer of a lightweight model. Then, we present a Progressive Meta-Pooling Learning (PMPL) strategy for the parameterized spatial enhancer to acquire a suitable receptive field size. The results on the ImageNet dataset demonstrate that MobileNetV2 using Meta-Pooling achieves top1 accuracy of 74.6\%, which outperforms MobileNetV2 by 2.3\%.