Customer retail environments represent an exciting and challenging context to develop and put in place cutting-edge computer vision techniques for more engaging customer experiences. Visual attention is one of the aspects that play such a critical role in the analysis of customers behaviour on advertising campaigns continuously displayed in shops and retail environments. In this paper, we approach the optimisation of advertisement layout content, aiming to grab the audience’s visual attention more effectively. We propose a fully automatic method for the delivery of the most effective layout content configuration using saliency maps out of each possible set of images with a given grid layout. Visual Saliency deals with the identification of the most critical regions out of pictures from a perceptual viewpoint. We want to assess the feasibility of saliency maps as a tool for the optimisation of advertisements considering all possible permutations of images which compose the advertising campaign itself. We start by analysing advertising campaigns consisting of a given spatial layout and a certain number of images. We run a deep learning-based saliency model over all permutations. Noticeable differences among global and local saliency maps occur over different layout content out of the same images. The latter aspect suggests that each image gives its contribution to the global visual saliency because of its content and location within the given layout. On top of this consideration, we employ some advertising images to set up a graphical campaign with a given design. We extract relative variance values out the local saliency maps of all permutations. We hypothesise that the inverse of relative variance can be used as an Effectiveness Score (ES) to catch those layout content permutations showing the more balanced spatial distribution of salient pixel. A group of 20 participants have run some eye-tracking sessions over the same advertising layouts to validate the proposed method.

A Saliency-Based Technique for Advertisement Layout Optimisation to Predict Customers’ Behaviour, 2021.

A Saliency-Based Technique for Advertisement Layout Optimisation to Predict Customers’ Behaviour

Alessandro Bruno
;
2021-01-01

Abstract

Customer retail environments represent an exciting and challenging context to develop and put in place cutting-edge computer vision techniques for more engaging customer experiences. Visual attention is one of the aspects that play such a critical role in the analysis of customers behaviour on advertising campaigns continuously displayed in shops and retail environments. In this paper, we approach the optimisation of advertisement layout content, aiming to grab the audience’s visual attention more effectively. We propose a fully automatic method for the delivery of the most effective layout content configuration using saliency maps out of each possible set of images with a given grid layout. Visual Saliency deals with the identification of the most critical regions out of pictures from a perceptual viewpoint. We want to assess the feasibility of saliency maps as a tool for the optimisation of advertisements considering all possible permutations of images which compose the advertising campaign itself. We start by analysing advertising campaigns consisting of a given spatial layout and a certain number of images. We run a deep learning-based saliency model over all permutations. Noticeable differences among global and local saliency maps occur over different layout content out of the same images. The latter aspect suggests that each image gives its contribution to the global visual saliency because of its content and location within the given layout. On top of this consideration, we employ some advertising images to set up a graphical campaign with a given design. We extract relative variance values out the local saliency maps of all permutations. We hypothesise that the inverse of relative variance can be used as an Effectiveness Score (ES) to catch those layout content permutations showing the more balanced spatial distribution of salient pixel. A group of 20 participants have run some eye-tracking sessions over the same advertising layouts to validate the proposed method.
Inglese
2021
https://link.springer.com/chapter/10.1007/978-3-030-68790-8_39
ICPR 2021: Pattern Recognition. ICPR International Workshops and Challenges
internazionale
contributo
ICPR 2021: Pattern Recognition. ICPR International Workshops and Challenges
495
507
978-3-030-68789-2
978-3-030-68790-8
Switzerland
Springer, Cham
esperti anonimi
Online
Settore INF/01 - Informatica
   Shoppar: Dynamically Optimised Digital Content.
   Innovate UK

   Shoppar: Dynamically Optimised Digital Content.
   Innovate UK
6
File in questo prodotto:
File Dimensione Formato  
Springer_Lecture_Notes_in_Computer_Science__Version_147_.pdf

Accessibile solo dagli utenti con account Apeiron

Tipologia: Documento in Post-print
Dimensione 1.38 MB
Formato Adobe PDF
1.38 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10808/50507
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 1
  • ???jsp.display-item.citation.isi??? ND
social impact