ARIGAN: Synthetic Arabidopsis Plants Using Generative Adversarial Network

Mario Valerio Giuffrida, Hanno Scharr, Sotirios A. Tsaftaris; Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2017, pp. 2064-2071

Abstract


In recent years, there has been an increasing interest in image-based plant phenotyping, applying state-of-the-art machine learning approaches. Despite the recent release of a few plant phenotyping datasets, large annotated plant image datasets are lacking. We propose an alternative solution to dataset augmentation for plant phenotyping, creating artificial images of plants using generative neural networks. We propose the Arabidopsis Rosette Image Generator (through) Adversarial Network: a deep convnet able to generate synthetic rosette plants, inspired by DCGAN. We trained the network using the CVPPP 2017 LCC dataset (only Arabidopsis). We show that our model generates realistic images of plants. We train our network conditioning on leaf count, such that it is possible to generate plants with a given number of leaves. Furthermore, we propose a new Ax dataset of artificial plants images, showing that the testing error is reduced when Ax is used as part of the training data.

Related Material


[pdf] [arXiv]
[bibtex]
@InProceedings{Giuffrida_2017_ICCV,
author = {Valerio Giuffrida, Mario and Scharr, Hanno and Tsaftaris, Sotirios A.},
title = {ARIGAN: Synthetic Arabidopsis Plants Using Generative Adversarial Network},
booktitle = {Proceedings of the IEEE International Conference on Computer Vision (ICCV) Workshops},
month = {Oct},
year = {2017}
}