Using Unlabeled Data for Increasing Low-Shot Classification Accuracy of Relevant and Open-Set Irrelevant Images
Keywords:
semi-supervised learning, open-set classification, neural networks, receiver operating characteristic
Abstract
In search exploration and reconnaissance tasks performed with autonomous ground vehicles an image classification capability is needed for specifically identifying targeted objects relevant classes and at the same time recognize when a candidate image does not belong to anyone of the relevant classes irrelevant images In this paper we present an open-set low-shot classifier that uses during its training a modest number less than 40 of labeled images for each relevant class and unlabeled irrelevant images that are randomly selected at each epoch of the training process The new classifier is capable of identifying images from the relevant classes determining when a candidate image is irrelevant and it can further recognize categories of irrelevant images that were not included in the training unseen The proposed low-shot classifier can be attached as a top layer to any pre-trained feature extractor when constructing a Convolutional Neural Network
Downloads
- Article PDF
- TEI XML Kaleidoscope (download in zip)* (Beta by AI)
- Lens* NISO JATS XML (Beta by AI)
- HTML Kaleidoscope* (Beta by AI)
- DBK XML Kaleidoscope (download in zip)* (Beta by AI)
- LaTeX pdf Kaleidoscope* (Beta by AI)
- EPUB Kaleidoscope* (Beta by AI)
- MD Kaleidoscope* (Beta by AI)
- FO Kaleidoscope* (Beta by AI)
- BIB Kaleidoscope* (Beta by AI)
- LaTeX Kaleidoscope* (Beta by AI)
How to Cite
Published
2022-05-26
Issue
Section
License
Copyright (c) 2022 Authors and Global Journals Private Limited
This work is licensed under a Creative Commons Attribution 4.0 International License.