Integrated Perception and Planning

This project considers the problem of how multi - modal and multi - sensor data acquisition and data analysis can be combined in a closed loop manner. This leads to more reliable decision making, while limiting the level of supervision needed. The project has the following over - arching goals.

Learning Multi-Agent, Multi-Modal Models with Limited Supervision
Reliable Decision-Making in Collaborative, Decentralized Team

Acknowledgment: ONR N00014-19-1-2264 from Science of AI program; previous grant NSF 1316934 from National Robotics Initiative

Learning with Limited Supervision

Sample Publications
- Exploiting Global Camera Network Constraints for Unsupervised Video Person Re-identification
  
  X. Wang, R. Panda, M. Liu, Y. Wang, A. Roy-Chowdhury, IEEE Trans. on Circuits and Systems for Video Technology (T-CSVT), 2020.
- Camera On-boarding for Person Re-identification using Hypothesis Transfer Learning
  
  Sk. M. Ahmed*, A. R. lejbolle*, R. Panda, A. Roy-Chowdhury, IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2020. (*joint first authors)
- Context-Aware Query Selection for Active Learning in Event Recognition [Supplemental Material]
  
  M. Hasan, S.Paul, A. Mourikis, A. Roy-Chowdhury, IEEE Trans. on Pattern Analysis and Machine Intelligence (T-PAMI), 2019.
- Non-Uniform Subset Selection for Active Learning in Structured Data
  
  S. Paul, J. H. Bappy, A. Roy-Chowdhury, IEEE Conf. on Computer Vision and Pattern Recognition, 2017.
- The Impact of Typicality for Informative Representative Selection
  
  J. H. Bappy, S. Paul, E. Tuncel and A. Roy-Chowdhury, IEEE Conf. on Computer Vision and Pattern Recognition, 2017.
- Continuous adaptation of multi-camera person identification models through sparse non-redundant representative selection [Supplemental Material]
  
  A. Das, R. Panda, A. Roy-Chowdhury, Computer Vision and Image Understanding, 2016.
- Online Adaptation for Joint Scene and Object Classification
  
  J. H. Bappy, S. Paul, A. Roy-Chowdhury, European Conf. on Computer Vision, 2016.
- Temporal Model Adaptation for Person Re-Identification
  
  N. Martinel, C. Micheloni, A. Roy-Chowdhury, European Conf. on Computer Vision, 2016.
- Generating Diverse Image Datasets with Limited Labeling
  
  N. C. Mithun, R. Panda, A. Roy-Chowdhury, ACM International Conf. on Multimedia, 2016.
- A Continuous Learning Framework for Activity Recognition Using Deep Hybrid Feature Models [Code]
  
  M. Hasan, A. Roy-Chowdhury, IEEE Trans. on Multimedia, 2015.

Robot Trajectory Learning

Sample Publications
- Learning from Trajectories via Subgoal Discovery [Code]
S. Paul, Jeroen van Baar, A. Roy-Chowdhury, Advances in Neural Information Processing Systems (NeurIPS), 2019

Vision and Learning Group

Integrated Perception and Planning

Learning with Limited Supervision

Robot Trajectory Learning

DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING