Python code for the cross-modal retrieval system proposed by N. Rasiwasia, J.C. Pereira, E. Coviello, G. Doyle, G.R.G. Lanckriet, R. Levy, N. Vasconcelos in the ACM MultiMedia '10 paper "A New Approach to Cross-Modal Multimedia Retrieval".

numpy sklean 0.16 scipy.io (to read .mat files)

• remove dependency on .mat format
• refactor code to run multiple experiments (CM, SM, SCM)

When using the code or the dataset, please cite:

@inproceedings{rasiwasia2010new,

title={A new approach to cross-modal multimedia retrieval},

author={Rasiwasia, Nikhil and Costa Pereira, Jose and Coviello, Emanuele and Doyle, Gabriel and Lanckriet, Gert RG and Levy, Roger and Vasconcelos, Nuno},

booktitle={Proceedings of the international conference on Multimedia},

pages={251--260},

year={2010},

organization={ACM}

}

@article{costa2014role,

title={On the role of correlation and abstraction in cross-modal multimedia retrieval},

author={Costa Pereira, Jose and Coviello, Emanuele and Doyle, Gabriel and Rasiwasia, Nikhil and Lanckriet, Gert RG and Levy, Roger and Vasconcelos, Nuno},

journal={Pattern Analysis and Machine Intelligence, IEEE Transactions on},

volume={36},

number={3},

pages={521--535},

year={2014},

publisher={IEEE}

}