DenseNet Models for Tiny ImageNet Classification

TL;DR

This paper introduces two DenseNet-based models tailored for Tiny ImageNet, utilizing specialized architecture, augmentation, and learning rate techniques to achieve competitive accuracy under limited resources.

Contribution

The paper develops two novel DenseNet architectures optimized for Tiny ImageNet, incorporating unique design choices and training techniques for resource-constrained environments.

Findings

Achieved ~60% top-1 validation accuracy

Utilized non-conventional augmentation and cyclical learning rates

Designed architectures based on dataset-specific image resolution

Abstract

In this paper, we present two image classification models on the Tiny ImageNet dataset. We built two very different networks from scratch based on the idea of Densely Connected Convolution Networks. The architecture of the networks is designed based on the image resolution of this specific dataset and by calculating the Receptive Field of the convolution layers. We also used some non-conventional techniques related to image augmentation and Cyclical Learning Rate to improve the accuracy of our models. The networks are trained under high constraints and low computation resources. We aimed to achieve top-1 validation accuracy of 60%; the results and error analysis are also presented.