A Comparative Study of Pre-Trained ConvNeXt and Vision Transformer Models for Image Classification

Honours Research Repository

[supervised by Dr Hairong Bau ([email protected])]

Stuart 'Lazarus' Groves - 2356823

For a full description and breakdown of the research pertaining to this repository please see the pdf inside the main directory. All references and usage credits are made there in full. For additional data from the research itself, please consult the Data.xlsx file. Three sheets contain a full collection of all the data that was obtained in order to reach the conclusion of the paper.

For the datasets, three datasets were used: 'LANDUSE', 'XRAY' and 'CIFAKE'

LANDUSE - https://www.kaggle.com/datasets/apollo2506/landuse-scene-classification (credit to user apollo2506)

XRAY - https://huggingface.co/datasets/keremberke/chest-xray-classification (credit to user keremberke)

CIFAKE - https://www.kaggle.com/datasets/birdy654/cifake-real-and-ai-generated-synthetic-images (credit to user birdy654)