Finding Beautiful and Happy Images for Mental Health and Well-being Applications

TL;DR

The paper tackles promoting mental health by automatically identifying beautiful and happy natural images. It introduces BNID, a large, high‑quality dataset with 20,996 images and about 420,000 human ratings, showing a meaningful correlation between beauty and happiness. A deep learning framework combines content-referenced image retrieval, Siamese feature comparison, and an emotion-assisted module to jointly predict beauty and happiness scores, with a final score $z(p)=z_{cr}(p)+z_{ea}(p)$. Experimental results demonstrate substantial improvements over baselines in predicting both beauty and happiness, supporting the feasibility of AI-powered image search to support mental well-being and AI-for-Good initiatives.

Abstract

This paper explores how artificial intelligence (AI) technology can contribute to achieve progress on good health and well-being, one of the United Nations' 17 Sustainable Development Goals. It is estimated that one in ten of the global population lived with a mental disorder. Inspired by studies showing that engaging and viewing beautiful natural images can make people feel happier and less stressful, lead to higher emotional well-being, and can even have therapeutic values, we explore how AI can help to promote mental health by developing automatic algorithms for finding beautiful and happy images. We first construct a large image database consisting of nearly 20K very high resolution colour photographs of natural scenes where each image is labelled with beautifulness and happiness scores by about 10 observers. Statistics of the database shows that there is a good correlation between the beautifulness and happiness scores which provides anecdotal evidence to corroborate that engaging beautiful natural images can potentially benefit mental well-being. Building on this unique database, the very first of its kind, we have developed a deep learning based model for automatically predicting the beautifulness and happiness scores of natural images. Experimental results are presented to show that it is possible to develop AI algorithms to automatically assess an image's beautifulness and happiness values which can in turn be used to develop applications for promoting mental health and well-being.

Figures (6)

• Figure 1: Left: Scatter plot between happiness and beautifulness scores. Right: Beautifulness and happiness score distributions
• Figure 2: Observers personal attributes and other factors versus raters' average beautifulness and happiness scores.
• Figure 3: The overall framework of our proposed algorithm for image beautifulness assessment. In all modules, the yellow boxes refer to the networks used, the brown and green boxes represent the features, in which the numbers indicate the channels of features, and the blue boxes represent the classifiers composed of fully connected layers. $\bm{p}$ is the input image and $\bm{q}$ is the paired reference image for comparison.
• Figure 4: Predicted beautifulness scores. The number below are predicted scores and ground truth scores (inside bracket)
• Figure 5: Example image and their predicted beautifulness-happiness difference. The numbers below each image represent predicted $beautifulness$ - $happiness$. A positive means the image's beauty score is higher than happiness score whilst a negative means means the image's happiness is higher than beautifulness.