AI reads emotions better when it sees bodies and surroundings

A new image system combines full-body posture with the surrounding scene instead of relying mainly on faces.

Editorial illustration — not from the study.

The researchers developed LEMOTIC, a dataset of 116,034 images showing a wide range of emotional situations. Their deep convolutional neural network used full-body posture and ambient information to recognize emotions in context, rather than focusing only on faces.

The system linked 26 emotional categories to three continuous dimensions: valence, or how pleasant an emotion is; arousal, or how activated it is; and dominance, or how much control a person appears to have. After tuning the model, the researchers reported better accuracy and generalization than earlier techniques for both types of emotional measures.

What the system recognized

The model recognized emotions from full-body postures and surrounding environmental information. It used the LEMOTIC dataset of 116,034 images and represented emotion through 26 categories as well as the dimensions of valence, arousal and dominance. The researchers report that the system outperformed earlier methods in accuracy and generalization for both discrete emotion categories and continuous emotional measures. Their results indicate that bodily and contextual information contributes to emotion recognition in images.

Why the whole scene matters

Images do not always show emotion clearly through a face alone. A person’s posture and surroundings can add information about what is happening, and the study provides an image-recognition approach designed to use those clues together. The researchers say this could support applications in social robotics, affective computing and human-computer interaction.

Evidence and caveats

This is a journal article describing a deep convolutional neural network trained and evaluated with the LEMOTIC dataset, after extensive hyperparameter tuning and comparison with earlier techniques. The abstract does not report the model’s numerical accuracy, the exact comparison methods, how the images were labeled, or results from testing on an independent dataset. It therefore supports the reported improvement within the study, but the abstract does not establish how well the system would perform across other image collections or real-world settings.

// Source

International Journal of Information Science and Technology · 2026 · DOI: 10.57675/imist.prsm/ijist-v10i1.382

View paper (DOI)Open access versionPublished 2026-10-04

Authors: Fatiha Limami, Boutaina Hdioud, Rachid Oulad Haj Thami

Tuna and billfish fishing moved offshore mainly in the past 200 years

People have fished for tuna for about 44,000 years and billfish for about 10,000 years, according to a global review of archaeological evidence. But sustained, intensive fishing in offshore and oceanic waters appears to have developed mainly during the past 200 years.

17 hours ago·5 views

Abstract editorial illustration accompanying the article: Coastal residents weigh flood damage, costs and family ties when considering a move

Society & Economics

Coastal residents weigh flood damage, costs and family ties when considering a move

People considering whether to leave flood-prone coastal communities weigh disastrous flooding and rising costs alongside crime, healthcare and proximity to family and friends. In a survey of 182 respondents, most said they would not relocate, while others said they would move immediately or in the future.

1 day ago·6 views

Society & Economics

Managers accepted reflective coaching from humans but wanted answers from AI

A study of eight managers found that they preferred human coaches for reflective questioning, even though they welcomed that approach from a person. When the same coaching protocol came from an AI system, they wanted it to provide answers instead.

1 day ago·23 views