Often multiple instances of an object occur in the same scene, for example in a warehouse. Unsupervised multi-instance object discovery algorithms are able to detect and identify such objects. We use such an algorithm to provide object proposals to a convolutional neural network (CNN) based classifier. This results in fewer regions to evaluate, compared to traditional region proposal algorithms. Additionally, it enables using the joint probability of multiple instances of an object, resulting in improved classification accuracy. The proposed technique can also split a single class into multiple sub-classes corresponding to the different object types, enabling hierarchical classification.
@article{arxiv.1707.07255,
title = {Detecting and Grouping Identical Objects for Region Proposal and Classification},
author = {Wim Abbeloos and Sergio Caccamo and Esra Ataer-Cansizoglu and Yuichi Taguchi and Chen Feng and Teng-Yok Lee},
journal= {arXiv preprint arXiv:1707.07255},
year = {2017}
}
Comments
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Workshop Deep Learning for Robotic Vision, 21 July, 2017, Honolulu, Hawaii