Object Localization with a Weakly Supervised CapsNet

• DOI: 10.48550/arxiv.1805.07706
• Publication Date: 2018-01-01
• Authors: Weitang Liu; Emad Barsoum; John D. Owens
• Keywords: MNIST database; Representation; Convolution (computer science)

Inspired by CapsNet's routing-by-agreement mechanism with its ability to learn object properties, we explore if those properties in turn can determine new of the objects, such as locations. We then propose a CapsNet architecture coordinate atoms and modified algorithm unevenly distributed initial routing probabilities. The model is based on but uses find objects' approximate positions image system. also discussed how derive property translation through show importance sparse representation. train our single moving MNIST dataset class labels. Our coordinates digits better than convolution counterpart that lacks algorithm, perform well when testing multi-digit KTH datasets. results method reaches state-of-art performance localization without any extra techniques modules prior work.