Brand-Attention

Abstract

In the highly competitive area of product marketing, the visibility of brand logos on packaging plays a crucial role in shaping consumer perception, directly influencing the success of the product. This paper introduces a comprehensive framework to measure the brand logo's attention on a packaging design. The proposed method consists of three steps. The first step leverages YOLOv8 for precise logo detection across prominent datasets, FoodLogoDet-1500 and LogoDet-3K. The second step involves modeling the user's visual attention with a novel saliency prediction model tailored for the packaging context. The proposed saliency model combines the visual elements with text maps employing a transformers-based architecture to predict user attention maps. In the third step, by integrating logo detection with saliency map generation, the framework provides a comprehensive brand attention score. The effectiveness of the proposed method is assessed module by module, ensuring a thorough evaluation of each component. Comparing logo detection and saliency map prediction with state-of-the-art models shows the superiority of the proposed methods. To investigate the robustness of the proposed brand attention score, we collected a unique dataset to examine previous psychophysical hypotheses related to brand visibility. Results show that the brand attention score is in line with all previous studies. Furthermore, we introduced seven new hypotheses to check the impact of position, orientation, presence of person, and other visual elements on brand attention. This research marks a significant stride in the intersection of cognitive psychology, computer vision, and marketing, paving the way for advanced, consumer-centric packaging designs.

Key Words: Brand Attention - Neuro Marketing - Logo Detection - Saliency Prediction

1. Logo-Detection

This module focuses on detecting brand logos in images using the YOLOv8 model. It utilizes two datasets for training: FoodLogoDet-1500 and LogoDet-3K.

❮ ❯

Table1: Metrics on Models Fine-Tuned over Foodlogo-det-1500

Method	mAP 50	mAP 50-95	Precision	Recall
MFDNet	0.879	0.635	0.836	0.811
YOLOv7	0.932	0.698	0.90	0.866
YOLOv8	0.936	0.704	0.904	0.879

Table2: Metrics on Models Pretrained on FoodLogo and Fine-Tuned over FoodLogoDet-1500+LogoDet3k Dataset

Method	mAP 50	mAP 50-95	Precision	Recall
MFDNet	0.87	0.62	0.82	0.8
YOLOv7	0.88	0.61	0.84	0.81
YOLOv8	0.94	0.71	0.91	0.88

2. Saliency-Map Prediction

This module is designed for predicting saliency maps of images, particularly suited for use in ads and packaging. The model leverages the ECSAL dataset for training. You can find the dataset here.

Table below shows ECT-SAL Performance against SOTA (State of the Art) methods (Average ± STDEV):

Method	CC ↑	KL ↓	AUC ↑	NSS ↑	SIM ↑
Contextual Encoder-Decoder	0.459 ± 0.136	1.13 ± 0.23	0.76 ± 0.066	0.925 ± 0.268	0.373 ± 0.06
DeepGazeIIE	0.561 ± 0.124	0.995 ± 0.215	0.842 ± 0.055	1.327 ± 0.318	0.399 ± 0.065
UNISAL	0.6 ± 0.15	0.768 ± 0.262	0.845 ± 0.056	1.574 ± 0.522	0.514 ± 0.094
EML-Net	0.510 ± 0.16	1.227 ± 0.903	0.807 ± 0.062	1.232 ± 0.407	0.536 ± 0.103
VGGSAM	0.691 ± 0.126	0.682 ± 0.259	0.815 ± 0.048	1.324 ± 0.362	0.58 ± 0.091
Transalnet	0.717 ± 0.061	0.873 ± 0.079	0.824 ± 0.054	1.723 ± 0.203	0.534 ± 0.043
VGGSSM	0.728 ± 0.121	0.599 ± 0.237	0.829 ± 0.043	1.396 ± 0.359	0.611 ± 0.089
Temp-SAL	0.719 ± 0.065	0.712 ± 0.126	0.813 ± 0.077	1.768 ± 0.182	0.629 ± 0.048
SSwin transformer	0.687 ± 0.175	0.652 ± 0.478	0.868 ± 0.072	1.701 ± 0.497	0.606 ± 0.101
Ours	0.75 ± 0.05	0.578 ± 0.117	0.892 ± 0.033	1.89 ± 0.204	0.645 ± 0.04

3. Brand Attention

The Brand Attention Module is a component designed to assess the visibility and attention of a brand within advertisement and packaging images. It combines logo detection and saliency map prediction techniques to quantify the presence and prominence of a brand in a given image.

Input Image

Brand-Attention Score: 23.54%

4. Object Attention

This Module is a component designed to assess the visibility and attention of any object you want within advertisement and packaging images. It saliency map prediction techniques to quantify the presence and prominence of that object in a given image.

Input Image

Object Attention Score:

BBox Selected

11.22%

In this section, we evaluate the effectiveness of the proposed brand attention module by comparing it with the observations in psychophysical studies. To test the model, we have designed a dataset where each group of images is the same in every way, apart from one particular logo feature it is examining. Wrapping up this section, we introduce some new hypotheses in this field that have not been explored yet.

1. Dataset

Our dataset comprises 650 images aimed at examining logo placement and packaging design's impact on brand perception. It integrates Internet-sourced images and AI-generated visuals, focusing on specific research queries through design alterations. The dataset is categorized into 12 hypotheses, with each set providing a controlled environment to study the distinct aspect of logo and packaging design influence.

Some images of the dataset generated for different theses and their brand attention scores are below:

2. Hypothesis

Our hypothesis and the results our model predicts are shown in the following table:

Table: Comparing the impact of packaging color and brand logo color on brand attention score

Hypothesis	Position	Mean	SE
Packaging Color	Black	36.82	5.65
	Brown	37.85	5.62
	Orange	37.46	5.46
	Yellow	37.45	5.61
	Green	36.38	5.55
	Blue	37.51	5.66
	Red	38.23	5.73
	White	40.84	5.89
Brand Logo Color	White	31.08	4.33
	Brown	34.54	4.4
	Orange	33.2	4.63
	Yellow	32.93	4.66
	Green	32.16	4.93
	Blue	33.13	4.87
	Black	36.56	4.7
	Red	37.44	4.79
Top-to-Bottom Logo Positioning	Down	28.89	5.19
	Up	34.05	5.64
	Center	40.02	7.06
All-Around Logo Positioning	Down-Right	15.05	3.05
	Down-Left	18.8	3.41
	Up-Right	16.51	3.05
	Up-Left	20.24	3.34
	Center	24.92	4.12
Bold Distinction	Boldness	19.98	2.27
Bold Distinction	Not Bold	21.1	2.35
Horizontal-Vertical Brand Logo Orientation	Horizontal	29.91	4.05
Horizontal-Vertical Brand Logo Orientation	Vertical	34.54	4.8
Horizontal-Vertical Packaging Orientation	Vertical	27.92	4.92
Horizontal-Vertical Packaging Orientation	Horizontal	36.92	5.59
Person in Packaging	With Person	32.26	5.76
Person in Packaging	No Person	36	6.16
Multi Object in Packaging	Multi	32.5	5
Multi Object in Packaging	One	40.95	5.29
Multi Packaging	Single	31.64	4.16
Multi Packaging	Multi	39.52	4.73
Text vs Image	Image	31.71	4.61
Text vs Image	Text	37.23	4.62
Square-Round Packaging Orientation	Round	25.82	4.86
Square-Round Packaging Orientation	Square	27.02	4.01