Learning Image Conditioned Label Space for Multilabel Classification.

RSS Source
Authors
Yi-Nan Li, Mei-Chen Yeh

This work addresses the task of multilabel image classification. Inspired bythe great success from deep convolutional neural networks (CNNs) forsingle-label visual-semantic embedding, we exploit extending these models formultilabel images. Specifically, we propose an image-dependent ranking model,which returns a ranked list of labels according to its relevance to the inputimage. In contrast to conventional CNN models that learn an imagerepresentation (i.e. the image embedding vector), the developed model learns amapping (i.e. a transformation matrix) from an image in an attempt todifferentiate between its relevant and irrelevant labels. Despite theconceptual simplicity of our approach, experimental results on a publicbenchmark dataset demonstrate that the proposed model achieves state-of-the-artperformance while using fewer training images than other multilabelclassification methods.

Stay in the loop.

Subscribe to our newsletter for a weekly update on the latest podcast, news, events, and jobs postings.