Deep Generalized Max Pooling

Christlein V, Spranger L, Seuret M, Nikolaou A, Král P, Maier A (2019)

Publication Language: English

Publication Type: Conference contribution

Publication year: 2019

Pages Range: 1090-1096

Conference Proceedings Title: 2019 International Conference on Document Analysis and Recognition (ICDAR)

Event location: Sydney AU

ISBN: 9781728130149


DOI: 10.1109/ICDAR.2019.00177


Global pooling layers are an essential part of Convolutional Neural Networks (CNN). They are used to aggregate activations of spatial locations to produce a fixed-size vector in several state-of-the-art CNNs. Global average pooling or global max pooling are commonly used for converting convolutional features of variable size images to a fix-sized embedding. However, both pooling layer types are computed spatially independent: each individual activation map is pooled and thus activations of different locations are pooled together. In contrast, we propose Deep Generalized Max Pooling that balances the contribution of all activations of a spatially coherent region by re-weighting all descriptors so that the impact of frequent and rare ones is equalized. We show that this layer is superior to both average and max pooling on the classification of Latin medieval manuscripts (CLAMM'16, CLAMM'17), as well as writer identification (Historical-WI'17).

Authors with CRIS profile

Involved external institutions

How to cite


Christlein, V., Spranger, L., Seuret, M., Nikolaou, A., Král, P., & Maier, A. (2019). Deep Generalized Max Pooling. In 2019 International Conference on Document Analysis and Recognition (ICDAR) (pp. 1090-1096). Sydney, AU.


Christlein, Vincent, et al. "Deep Generalized Max Pooling." Proceedings of the 2019 International Conference on Document Analysis and Recognition (ICDAR), Sydney 2019. 1090-1096.

BibTeX: Download