1 / 28

Geometry-Preserving Visual Phrases for Enhanced Image Retrieval

This paper discusses the innovative approach of using geometry-preserving visual phrases to improve image retrieval systems. Traditional methods rely on bag-of-words, which often neglect geometric modeling, leading to inefficiencies in search relevance. Our proposed model effectively combines extensive geometric analysis while maintaining computational efficiency. We present experiments conducted on the Flicker 1M dataset, highlighting significant improvements in precision and recall compared to conventional techniques. The results demonstrate the effectiveness of our approach in addressing intrinsic challenges in image retrieval.

rayya
Télécharger la présentation

Geometry-Preserving Visual Phrases for Enhanced Image Retrieval

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Image Retrieval with Geometry-Preserving Visual Phrases Yimeng Zhang ZhaoyinJiaTsuhan Chen School of Electrical and Computer Engineering, Cornell University

  2. OUTLINE • Introduction • GVP • Experiments • Conclusion

  3. OUTLINE • Introduction • GVP • Experiments • Conclusion

  4. Image Retrieval Image Database Ranked relevant images and metadata …

  5. Challenges

  6. Bag of Words

  7. Pros and cons • Pros • Computationally efficient • Cons • No shape/geometry modeling

  8. Phrases vs. Words

  9. Previous work

  10. Goal To model unbounded order features with extensive geometry modeling, but same computational complexity with bag of words

  11. Dataset

  12. OUTLINE • Introduction • GVP • Experiments • Conclusion

  13. Mutual word relationship

  14. Correspondence transform

  15. Correspondence transform

  16. Inverted Index with BoW

  17. Inverted Index with Phrases

  18. Inverted Index with Phrases

  19. Final Score

  20. Increase the invariance

  21. OUTLINE • Introduction • GVP • Experiments • Conclusion

  22. Example Precision-recall curve

  23. Comparison

  24. Flicker 1M dataset

  25. OUTLINE • Introduction • GVP • Experiments • Conclusion

  26. Conclusions

  27. THANK YOU

More Related