Optimizing Flower Classification Models: A Comparative Analysis of Bayesian Optimization and Random Search for Hyper parameter Tuning

DOI

10.2991/978-94-6463-716-8_16How to use a DOI?

Keywords

Flower Classification; Swim Transformer; Random Search; Bayesian Search; Hyper Parameter

Abstract

Understanding plant varieties, ecosystem health, and agricultural methods are made easier with the help of flower categorisation, which is essential to botanical study, environmental monitoring, and agriculture. The effectiveness of automating flower classification has been greatly attributed to recent developments in machine learning, particularly in deep neural networks. But in order to maximise model performance, hyper parameter optimisation is crucial. By utilising state-of-the-art hyper parameter optimisation approaches, this research seeks to improve the precision and efficacy of floral categorisation systems. The work addresses issues such dataset anomalies, such as photos of flowers in unusual locations, by utilising a Kaggle dataset with 104 flower species. In order to maximise the macro F1 score—a crucial metric for multiclass classification—the study contrasts random search versus Bayesian optimisation for hyper parameter tweaking. The study obtained an F1 score of 0.966 with random search using the Swin Transformer model, as opposed to 0.9285 with Bayesian optimisation. With batch sizes optimised for TPU utilisation, the dataset was divided into 77.46% for training and 22.54% for validation.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Naresh Dembla
AU  - Ravindra Yadav
AU  - Urvashi Sharma
AU  - Devendra Singh
PY  - 2025
DA  - 2025/05/26
TI  - Optimizing Flower Classification Models: A Comparative Analysis of Bayesian Optimization and Random Search for Hyper parameter Tuning
BT  - Proceedings of the International Conference on Recent Advancements and Modernisations in Sustainable Intelligent Technologies and Applications (RAMSITA 2025)
PB  - Atlantis Press
SP  - 189
EP  - 199
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-716-8_16
DO  - 10.2991/978-94-6463-716-8_16
ID  - Dembla2025
ER  -