Using Unlabeled Data for Increasing Low-Shot Classification Accuracy of Relevant and Open-Set Irrelevant Images
This addresses the need for robust image classification in autonomous search and reconnaissance tasks, though it is incremental as it builds on existing pre-trained feature extractors.
The paper tackles the problem of open-set low-shot classification for autonomous ground vehicles, using less than 40 labeled images per relevant class and unlabeled irrelevant images to achieve accurate identification of relevant objects and recognition of irrelevant images, including unseen categories.
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.