TR2017-099

Detecting and Grouping Identical Objects for Region Proposal and Classification

    •  Abbeloos, W., Caccamo, S., Ataer-Cansizoglu, E., Taguchi, Y., Feng, C., Lee, T.-Y., "Detecting and Grouping Identical Objects for Region Proposal and Classification", CVPR Workshop on Deep Learning for Robotic Vision, DOI: 10.1109/​CVPRW.2017.76, July 2017.
      BibTeX TR2017-099 PDF
      • @inproceedings{Abbeloos2017jul,
      • author = {Abbeloos, Wim and Caccamo, Sergio and Ataer-Cansizoglu, Esra and Taguchi, Yuichi and Feng, Chen and Lee, Teng-Yok},
      • title = {Detecting and Grouping Identical Objects for Region Proposal and Classification},
      • booktitle = {CVPR Workshop on Deep Learning for Robotic Vision},
      • year = 2017,
      • month = jul,
      • doi = {10.1109/CVPRW.2017.76},
      • url = {https://www.merl.com/publications/TR2017-099}
      • }
  • Research Areas:

    Artificial Intelligence, Computer Vision, Machine Learning, Robotics

Abstract:

Often multiple instances of an object occur in the same scene, for example in a warehouse. Unsupervised multiinstance object discovery algorithms are able to detect and identify such objects. We use such an algorithm to provide object proposals to a CNN-based classifier. This results in fewer regions to evaluate, compared to traditional region proposal algorithms. Additionally, it enables using the joint probability of multiple instances of an object, resulting in improved classification accuracy. The proposed technique can also split a single class into multiple sub-classes corresponding to the different object types, enabling hierarchical classification.