Work

Facial Expression Classification in The Wild by Traditional Machine Learning and Deep Learning

Year: 2025
Type: article
Abstract: Analyzing facial expressions is a fundamental aspect of computer vision, with significant importance in applications such as human-machine communication and interaction. Recognizing expressions in uncontrolled environments (wild), with varied lighting, poses, and occlusions, is particularly challenging. This study compares traditional methods (PCA, LBP, LDA, and HOG) with deep learning technique (CNN) using CK+ (posed images) and FER2013 (candid images) datasets. Results show that CNN outperforms traditional methods, achieving 34% higher accuracy in uncontrolled conditions. While traditional approaches excel in controlled settings, deep learning proves more effective overall for FER in natural environments
Language: English
Open Access status: diamond
OpenAlex ID: W7219529338
FWCI:
Cites: 72
Cited by:
Related to: 10