Tampere University of Technology

TUTCRIS Research Portal

Object Proposals using CNN-based edge filtering

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Standard

Object Proposals using CNN-based edge filtering. / Waris, Muhammad Adeel; Iosifidis, Alexandros; Gabbouj, Moncef.

2016 23rd International Conference on Pattern Recognition (ICPR). IEEE, 2017.

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Harvard

Waris, MA, Iosifidis, A & Gabbouj, M 2017, Object Proposals using CNN-based edge filtering. in 2016 23rd International Conference on Pattern Recognition (ICPR). IEEE, International Conference on Pattern Recognition, 1/01/00. https://doi.org/10.1109/ICPR.2016.7899704

APA

Waris, M. A., Iosifidis, A., & Gabbouj, M. (2017). Object Proposals using CNN-based edge filtering. In 2016 23rd International Conference on Pattern Recognition (ICPR) IEEE. https://doi.org/10.1109/ICPR.2016.7899704

Vancouver

Waris MA, Iosifidis A, Gabbouj M. Object Proposals using CNN-based edge filtering. In 2016 23rd International Conference on Pattern Recognition (ICPR). IEEE. 2017 https://doi.org/10.1109/ICPR.2016.7899704

Author

Waris, Muhammad Adeel ; Iosifidis, Alexandros ; Gabbouj, Moncef. / Object Proposals using CNN-based edge filtering. 2016 23rd International Conference on Pattern Recognition (ICPR). IEEE, 2017.

Bibtex - Download

@inproceedings{3c5cb91e5f044cc6b737ebeba2452a86,
title = "Object Proposals using CNN-based edge filtering",
abstract = "With the success of deep learning in the last few years, the object detection community shifted from processing on exhaustive sliding windows to smaller set of object proposals using more powerful and deep visual representations. Object proposals increase the accuracy and speed up detection process by reducing the search space. In this paper we propose a novel idea of filtering irrelevant edges using semantic image filtering and true objectness learnt within convolutional layers of CNN. Our approach localizes well proposals by producing highly accurate bounding boxes and reduces the number of proposals. The greatest benefit of our approach is that it can be integrated into any existing method exploiting edge-based objectness to achieve consistently high recall across various intersection over union thresholds. Unlike other supervised methods, our approach does not require bounding box annotations for training. Experiments on PASCAL VOC 2007 dataset demonstrate that our approach improves the state-of-the-art model with a significant margin.",
author = "Waris, {Muhammad Adeel} and Alexandros Iosifidis and Moncef Gabbouj",
year = "2017",
doi = "10.1109/ICPR.2016.7899704",
language = "English",
booktitle = "2016 23rd International Conference on Pattern Recognition (ICPR)",
publisher = "IEEE",

}

RIS (suitable for import to EndNote) - Download

TY - GEN

T1 - Object Proposals using CNN-based edge filtering

AU - Waris, Muhammad Adeel

AU - Iosifidis, Alexandros

AU - Gabbouj, Moncef

PY - 2017

Y1 - 2017

N2 - With the success of deep learning in the last few years, the object detection community shifted from processing on exhaustive sliding windows to smaller set of object proposals using more powerful and deep visual representations. Object proposals increase the accuracy and speed up detection process by reducing the search space. In this paper we propose a novel idea of filtering irrelevant edges using semantic image filtering and true objectness learnt within convolutional layers of CNN. Our approach localizes well proposals by producing highly accurate bounding boxes and reduces the number of proposals. The greatest benefit of our approach is that it can be integrated into any existing method exploiting edge-based objectness to achieve consistently high recall across various intersection over union thresholds. Unlike other supervised methods, our approach does not require bounding box annotations for training. Experiments on PASCAL VOC 2007 dataset demonstrate that our approach improves the state-of-the-art model with a significant margin.

AB - With the success of deep learning in the last few years, the object detection community shifted from processing on exhaustive sliding windows to smaller set of object proposals using more powerful and deep visual representations. Object proposals increase the accuracy and speed up detection process by reducing the search space. In this paper we propose a novel idea of filtering irrelevant edges using semantic image filtering and true objectness learnt within convolutional layers of CNN. Our approach localizes well proposals by producing highly accurate bounding boxes and reduces the number of proposals. The greatest benefit of our approach is that it can be integrated into any existing method exploiting edge-based objectness to achieve consistently high recall across various intersection over union thresholds. Unlike other supervised methods, our approach does not require bounding box annotations for training. Experiments on PASCAL VOC 2007 dataset demonstrate that our approach improves the state-of-the-art model with a significant margin.

U2 - 10.1109/ICPR.2016.7899704

DO - 10.1109/ICPR.2016.7899704

M3 - Conference contribution

BT - 2016 23rd International Conference on Pattern Recognition (ICPR)

PB - IEEE

ER -