Deep Network Pruning for Object Detection
Ghosh S, Srinivasa SK, Amon P, Hutter A, Kaup A (2019)
Publication Type: Conference contribution
Publication year: 2019
Journal
Publisher: IEEE Computer Society
Book Volume: 2019-September
Pages Range: 3915-3919
Conference Proceedings Title: Proceedings - International Conference on Image Processing, ICIP
Event location: Taipei
ISBN: 9781538662496
DOI: 10.1109/ICIP.2019.8803505
Abstract
With the increasing success of deep learning in various applications, there is an increasing need to have deep models that can be used for deployment in real-time and/or resource constrained scenarios. In this context, this paper analyzes the pruning of deep models for object detection in order to reduce the number of weights and hence the number of computations. Very deep networks based on ResNet like architectures, like YOLOv3 have unique challenges when attempting to prune them. This paper proposes a network pruning technique based on agglomerative clustering for the feature extractor and using mutual information for the detector. The performance of the proposed techniques is also compared with that of a relatively shallow network, i.e., YOLOv2. A compression percentage of around 30% results in a 10% drop of mean average precision (mAP) in YOLOv3, whereas in YOLOv2 the drop was around 6% on the COCO dataset.
Authors with CRIS profile
Sanjukta Ghosh
Lehrstuhl für Multimediakommunikation und Signalverarbeitung
Andre Kaup
Lehrstuhl für Multimediakommunikation und Signalverarbeitung
How to cite
APA:
Ghosh, S., Srinivasa, S.K., Amon, P., Hutter, A., & Kaup, A. (2019). Deep Network Pruning for Object Detection. In Proceedings - International Conference on Image Processing, ICIP (pp. 3915-3919). Taipei, TW: IEEE Computer Society.
MLA:
Ghosh, Sanjukta, et al. "Deep Network Pruning for Object Detection." Proceedings of the 26th IEEE International Conference on Image Processing, ICIP 2019, Taipei IEEE Computer Society, 2019. 3915-3919.
BibTeX: Download