April 24, 2017

Taxonomy augmented object recognition

Realistic scene object recognition in computer vision still faces great challenges due to the large intra-class variation of object images caused by factors like object appearance variation and viewpoint change. To address this challenge, we propose to exploit the semantic relationships embedded in object taxonomy for improved object recognition. Specifically, we exploit the relationships in the object taxonomy to augment the learning of object classifiers. We utilize two types of relationships in the taxonomy, including the overall relationship and the local relationship. Our proposed approach jointly incorporates both the overall relationship as the loss function for classifier learning, and the local relationship as classifier learning constraints. Experiments on benchmark datasets demonstrate the effectiveness of our method in incorporating taxonomy for object recognition compared to the state-art-the-art methods.

View Original Article

Recent Publications

June 04, 2017

A New PRACH Transmission Scheme in Unlicensed Spectrum

  • Luo Z.
  • Meng Y.
  • Tao T.

For the unlicensed spectrum, the occupied bandwidth requirement is demanded by some regulations. The legacy scheme of Physical Random Access Channel (PRACH) for Long Term Evolution (LTE) cannot satisfy it. In this paper, we propose a novel PRACH transmission scheme to satisfy the requirement of unlicensed spectrum based on preamble ...

June 01, 2017

Mutual service processes in Euclidean spaces: existence and ergodicity

  • Baccelli F.
  • Mathieu F.
  • Norros I.

Consider a set of objects, abstracted to points of a spatially stationary point process in R-d, that deliver to each other a service at a rate depending on their distance. Assume that the points arrive as a Poisson process and leave when their service requirements have been fulfilled. We show ...

June 01, 2017

Incentivizing social media users for mobile crowdsourcing

  • Aiello L.
  • Karaliopoulos M.
  • Koutsopoulos I.
  • Micholia P.
  • Morales G.
  • Quercia D.

We focus on the problem of contributor-task matching in mobile crowd-sourcing. The idea is to identify existing social media users who possess domain expertise (e.g., photography) and incentivize them to perform some tasks (e.g., take quality pictures). To this end, we propose a framework that extracts the potential contributors' expertise ...