arXiv reaDer
Classification of freshwater snails of the genus Radomaniola with multimodal triplet networks
In this paper, we present our first proposal of a machine learning system for the classification of freshwater snails of the genus Radomaniola. We elaborate on the specific challenges encountered during system design, and how we tackled them; namely a small, very imbalanced dataset with a high number of classes and high visual similarity between classes. We then show how we employed triplet networks and the multiple input modalities of images, measurements, and genetic information to overcome these challenges and reach a performance comparable to that of a trained domain expert.
updated: Tue Jul 30 2024 08:56:25 GMT+0000 (UTC)
published: Mon Jul 29 2024 13:45:23 GMT+0000 (UTC)
参考文献 (このサイトで利用可能なもの) / References (only if available on this site)
被参照文献 (このサイトで利用可能なものを新しい順に) / Citations (only if available on this site, in order of most recent)
Amazon.co.jpアソシエイト