Comparative Analysis of Lightweight Deep Learning Models for Memory-Constrained Devices

TL;DR

This paper evaluates lightweight deep learning models for image classification on resource-limited devices, comparing their accuracy, efficiency, and suitability for deployment in edge computing environments.

Contribution

It provides a comprehensive benchmark of five state-of-the-art lightweight models across multiple datasets, analyzing the effects of transfer learning, hyperparameter tuning, and training methods.

Findings

EfficientNetV2 achieves the highest accuracy among models.

MobileNetV3 offers the best balance between accuracy and efficiency.

SqueezeNet excels in inference speed and compactness.

Abstract

This paper presents a comprehensive evaluation of lightweight deep learning models for image classification, emphasizing their suitability for deployment in resource-constrained environments such as low-memory devices. Five state-of-the-art architectures - MobileNetV3 Small, ResNet18, SqueezeNet, EfficientNetV2-S, and ShuffleNetV2 - are benchmarked across three diverse datasets: CIFAR-10, CIFAR-100, and Tiny ImageNet. The models are assessed using four key performance metrics: classification accuracy, inference time, floating-point operations (FLOPs), and model size. Additionally, we investigate the impact of hyperparameter tuning, data augmentation, and training paradigms by comparing pretrained models with scratch-trained counterparts, focusing on MobileNetV3 Small. Our findings reveal that transfer learning significantly enhances model accuracy and computational efficiency, particularly for complex datasets like Tiny ImageNet. EfficientNetV2 consistently achieves the highest accuracy, while MobileNetV3 offers the best balance between accuracy and efficiency, and SqueezeNet excels in inference speed and compactness. This study highlights critical trade-offs between accuracy and efficiency, offering actionable insights for deploying lightweight models in real-world applications where computational resources are limited. By addressing these challenges, this research contributes to optimizing deep learning systems for edge computing and mobile platforms.