Bag of Surrogate Parts: one inherent feature of deep CNNs

Yanming Guo and Michael S. Lew

Abstract

Convolutional Neural Networks (CNNs) have achieved promising performance in image classification tasks. In this paper, we develop a new feature from convolutional layers, called Bag of Surrogate Parts (BoSP), and its spatial variant, Spatial BoSP (S-BoSP). Specifically, we take the feature maps in convolutional layers as surrogate parts, and densely sample and assign the regions in input images to these surrogate parts by observing the activation values. To better handle the objects with different scales and deformations, and make more comprehensive predictions, we further propose a scale pooling technique for assigning the features, and global constrained augmentation for the final prediction. Compared with most existing methods that also utilize the activations from convolutional layers, the proposed method is efficient, has no tuning parameters, and could generate low-dimensional, highly discriminative features. The experiments on generic object, fine-grained object and scene datasets indicate that the proposed feature can not only produce superior results to fully-connected layer based features, but also get comparable, or in some cases considerably better performance than the state-of-the-art.

Session

Posters 2

Files

Extended Abstract (PDF, 180K)
Paper (PDF, 308K)

DOI

10.5244/C.30.96
https://dx.doi.org/10.5244/C.30.96

Citation

Bibtex

        @inproceedings{BMVC2016_96,
        	title={Bag of Surrogate Parts: one inherent feature of deep CNNs},
        	author={Yanming Guo and Michael S. Lew},
        	year={2016},
        	month={September},
        	pages={96.1-96.12},
        	articleno={96},
        	numpages={12},
        	booktitle={Proceedings of the British Machine Vision Conference (BMVC)},
        	publisher={BMVA Press},
        	editor={Richard C. Wilson, Edwin R. Hancock and William A. P. Smith},
        	doi={10.5244/C.30.96},
        	isbn={1-901725-59-6},
        	url={https://dx.doi.org/10.5244/C.30.96}
        }