MergedNET: A simple approach for one-shot learning in siamese networks based on similarity layers

• DOI: 10.1016/j.neucom.2022.08.070
• Publication Date: 2022-08-17T15:11:12Z
• Authors: John Atanbori; Samuel Rose
• Keywords: Similarity (geometry); Single shot; One shot

Classifiers trained on disjointed classes with few labelled data points are used in one-shot learning to identify visual concepts from other classes. Recently, Siamese networks and similarity layers have been solve the problem, achieving state-of-the-art performance visual-character recognition datasets. Various techniques developed over years improve of these fine-grained image classification They focused primarily improving loss activation functions, augmenting features, employing multiscale metric learning, pre-training fine-tuning backbone network. We investigate for tasks propose two frameworks combining into a MergedNet On all four datasets our experiment, outperformed baselines based accuracy, it generalises when miniImageNet.