Genuine product appreciation using automated facial expression recognition
Date of Publication
Master of Science in Computer Science
College of Computer Studies
Merlin Teodosia C. Suarez
Products have been designed and manufactured to satisfy customer needs. Market research uses survey to gather customers feedback but sometimes customers need to answer tedious surveys and can influence other customers. This work proposes the use of automatic facial expression analysis to gather customer feedback. The input will be full frontal face video and the system will use Constrained Local Model uses three algorithms, active shape model (ASM), active appearance model (AAM) and pictorial structure matching (PSM), to detect and extract the features of the face. After extracting the features, the system will classify the emotions based on the features and will be tested through real-time video. The corpus contains 160 facial distance features with emotion labels and 267,824 instances of 55 subjects. The given emotion labels are happy, satisfaction, others, enthusiastic, neutral, interest, disgust, and appreciation. Principal component analysis (PCA) and correlation-based feature selection (CFS) were used for feature selection to determine the best features. C4.5 and support vector machine (SVM) with gaussian kernel function were used to classify the emotion. The results showed that C4.5 has the highest accuracy with 76.23%. Comparing the classifiers with feature selection, C4.5 has higher accuracy with 74.96% for CFS and 68.08% for PCA than SVM with 74.51% for CFS and 56.27% for PCA. Keywords: Image Processing, Constrained Local Model, Product Appreciation, Emotion Recognition, Genuine Feedback, Correlation-based feature selection, C4.5, gaussian kernel, principal component analysis, support vector machine.
Archives, The Learning Commons, 12F Henry Sy Sr. Hall
1 v. (various foliations) ; c28 cm. + 1 computer optical disc.
Image processing; Emotion recognition; Support vector machines; Market surveys
Choi, E. (2012). Genuine product appreciation using automated facial expression recognition. Retrieved from https://animorepository.dlsu.edu.ph/etd_masteral/4350