Classification of Rich-Classes but Scarce-Samples Images via Multi-modeling: the Humpback Whale Epitome
Haojiong Wang, Hiroshi Tsutsui, Matteo Convertino · 2022 IEEE 4th Global Conference on Life Sciences and Technologies (LifeTech) · 2022
This paper focuses on demonstrating classification techniques for image datasets with rich classes but scarce samples per category. Although the machine learning technique is an effective way to solve such problems, it is difficult to define a particular method as optimal among traditional machine learning methods. We propose the combination of multiple Siamese Networks with a voting model for the classification of such complex datasets. Experimental results show that our approach can outperform the results of several traditional Siamese classification models working independently when dealing with rich classes but sparsely sampled image data.