Multi-task CNN for multi-cue affects recognition using upper-body gestures and facial expressions

Zaghbani, Soumaya; Bouhlel, Med Salim

doi:10.1007/s41870-021-00820-w

Cited by 5 publications

(1 citation statement)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This has allowed training even the most complex neural networks in terms of architecture, opening up a whole horizon of previously unsolvable tasks [1][2]. Currently, intelligent systems are not only focused on image recognition but also on extracting metadata, such as emotions, mood, gender or age of a person, from already recognized objects [3][4]. Emotions play an important role in human life and interpersonal communication [5] and represent a psychological state of the human mind.…”

Section: Introductionmentioning

confidence: 99%

Development of an emotion and age recognition model based on the combination of local binary patterns and squeeze-and-excitation block method

Soma,

Kopanitsa

2024

Third International Conference on Digital Technologies, Optics, and Materials Science (DTIEE 2024)

View full text Add to dashboard Cite

This paper presents models for improving accuracy and efficiency in recognizing a person's emotional state and age based on facial expressions using the combination of Local Binary Patterns and Squeeze-and-Excitation Block methods. The proposed models were trained on the AffectNet dataset for emotion recognition and the Adience dataset for age recognition with four classes [angry, happy, neutral, sad] and [6-20, 25-30, 42-48, 60-98]. The accuracy and speed of these models, achieved after training, were compared with the results described in scientific articles published in the Scopus database. According to the obtained experimental data, the first and second developed models achieved validation accuracy of 81.5% and 72.3%, respectively.

show abstract