Work

Pollen image manipulation and projection using latent space

Year: 2025

Type: article

Abstract: Understanding the structure of pollen grains is crucial for the identification of plant taxa and the understanding of plant evolution. We employ a deep learning technique known as style transfer to investigate the manipulation of microscope images of these pollens to change the size and shape of pollen grain images. This methodology unveils the potential to identify distinctive structural features of pollen grains and decipher correlations, whilst the ability to generate images of pollen can enhance our capacity to analyse a larger variety of pollen types, thereby broadening our understanding of plant ecology. This could potentially lead to advancements in fields such as agriculture, botany, and climate science.

Source: Frontiers in Plant Science

DOI: 10.3389/fpls.2025.1539128

Authors: Ben Mills, Michalis N. Zervas, James A. Grant-Jacob

Institution: University of Southampton

Language: English

Topic: Generative Adversarial Networks and Image Synthesis

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $3,801

Funder: Engineering and Physical Sciences Research Council