2021 IEEE International Conference on Image Processing

19-22 September 2021 • Anchorage, Alaska, USA

Imaging Without Borders

2021 IEEE International Conference on Image Processing

19-22 September 2021 • Anchorage, Alaska, USA

Login Paper Search My Schedule Paper Index Help

My ICIP 2021 Schedule

Note: Your custom schedule will not be saved unless you create a new account or login to an existing account.
  1. Create a login based on your email (takes less than one minute)
  2. Perform 'Paper Search'
  3. Select papers that you desire to save in your personalized schedule
  4. Click on 'My Schedule' to see the current list of selected papers
  5. Click on 'Printable Version' to create a separate window suitable for printing (the header and menu will appear, but will not actually print)

Paper Detail

Paper IDMLR-APPL-IP-3.4
Paper Title Multiscale IoU: A Metric for Evaluation of Salient Object Detection with Fine Structures
Authors Azim Ahmadzadeh, Dustin J. Kempton, Yang Chen, Rafal A. Angryk, Georgia State University, United States
SessionMLR-APPL-IP-3: Machine learning for image processing 3
LocationArea F
Session Time:Tuesday, 21 September, 08:00 - 09:30
Presentation Time:Tuesday, 21 September, 08:00 - 09:30
Presentation Poster
Topic Applications of Machine Learning: Machine learning for image processing
IEEE Xplore Open Preview  Click here to view in IEEE Xplore
Abstract Increasingly, general-purpose object-detection algorithms are being applied to detection tasks of singular focus, where their generality poses some limitations. The coarse segmentation of objects is one of these limitations, and can be traced back to how we evaluate their detection precision. Recognizing the value of this generality, and the limited use of task-specific algorithms, our goal is to re-negotiate this trade-off and close the gap between these two worlds. In this work, we present a new metric that is a marriage of a popular evaluation metric, named Intersection over Union (IoU), and fractal dimension. Using the ideas behind these concepts, we propose Multiscale IoU (MIoU) which allows comparison of the detected and ground-truth regions at multiple resolution levels. Through several examples, we show that MIoU is indeed sensitive to the fine boundary structures which can be completely overlooked by other region-based metrics. We further examine the reliability of MIoU using synthetic and real-world datasets of objects, and show that its values follow the same distribution as those of IoU do. We intend this work (with reproducible experiments) to re-initiate exploration of new evaluation methods for object-detection algorithms.