A Geometry-Sensitive Approach for Photographic Style Classification

Abstract

Photographs are characterized by different compositional attributes like the Rule of Thirds, depth of field, vanishing-lines etc. The presence or absence of one or more of these attributes contributes to the overall artistic value of an image. In this work, we analyze the ability of deep learning based methods to learn such photographic style attributes. We observe that although a standard CNN learns the texture and appearance based features reasonably well, its understanding of global and geometric features is limited by two factors. First, the data-augmentation strategies (cropping, warping, etc.) distort the composition of a photograph and affect the performance. Secondly, the CNN features, in principle, are translation-invariant and appearance-dependent. But some geometric properties important for aesthetics, e.g. the Rule of Thirds (RoT), are position-dependent and appearance-invariant. Therefore, we propose a novel input representation which is geometry-sensitive, position-cognizant and appearance-invariant. We further introduce a two-column CNN architecture that performs better than the state-of-the-art (SoA) in photographic style classification. From our results, we observe that the proposed network learns both the geometric and appearance-based attributes better than the SoA.

Collaborators : Koustav Ghosal, Mukta Prasad, Aljosa Smolic

Code : https://github.com/V-Sense/A-Geometry-Sensitive-Approach-for-photographic-Style-Classification.git

Paper : https://v-sense.scss.tcd.ie/wp-content/uploads/2018/08/IMVIP_2018_paper_2-2.pdf

Slides : IMVIP_2018_KOUSTAV_SLIDES

This entry was posted in Visual Effects (VFX) and tagged , , , by Koustav Ghosal. Bookmark the permalink.

About Koustav Ghosal

Koustav is a PhD student with the V-SENSE project working in the area of developing intelligent systems for aesthetic prediction of multimedia. His area of interest also includes Computational Photography, Action Recognition, Multimedia Retrieval and Classification. Before joining the V-SENSE group, Koustav pursued MS (by research) in Computer Science from Centre of Visual Information Technology, International Institute of Information Technology, Hyderabad, India. During master's his area of research was sketch-based multimedia retrieval.