A fusion of CNN And SIFT For multicultural facial expression recognition

• Published in: Multimedia tools and applications Vol. 84; no. 28; pp. 33505 - 33523
• Main Authors: Qadir, Inam, Iqbal, Muhammad Amjad, Ashraf, Samman, Akram, Sheeraz
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.08.2025; Springer Nature B.V
• Subjects: Accuracy; Algorithms; Artificial intelligence; Classification; Computer Communication Networks; Computer Science; Data Structures and Information Theory; Datasets; Emotions; Face recognition; Grafting; Machine learning; Multiculturalism & pluralism; Multimedia Information Systems; Neural networks; Pattern recognition; Special Purpose and Application-Based Systems; Support vector machines; White people; Facial expression recognition; CNN; Multicultural; Feature grafting; Cross cultural
• ISSN: 1573-7721; 1380-7501; 1573-7721
• DOI: 10.1007/s11042-024-20589-x

The most generic and understandable way of communication is by observing facial expressions; Facial Expression Recognition(FER) performance was affected by the differences in ethnicity, culture, and geography. This research proposes a feature grafting-based novel technique to recognize multicultural facial expressions. The Viola-Jones algorithm detects the facial region of an image that is most concentrated. Feature Grafting is used to combine high-level and low-level features. A multicultural facial expression database was created using still images of male and female subjects obtained from three datasets: TFEID, Radboud, and JAFFE. The model is designed and tested using this dataset and the proposed technique. The obtained accuracy using the proposed technique was 94.34% which is better than the state-of-the-art method.