Font Size: a A A

The Application Research Of Facial Expression Analysis Method In Usability Evaluation

Posted on:2016-05-10Degree:MasterType:Thesis
Country:ChinaCandidate:C YangFull Text:PDF
GTID:2308330461465038Subject:Applied Psychology
Abstract/Summary:PDF Full Text Request
Facial expression analysis method in usability evaluation refers to direct measure and analysis of the expressions generated in the process of using the product. With the direct and real-time advantage, facial expression analysis makes up for the traditional usability methods’ indirect, hysteretic disadvantage. Hence, more and more researchers began to pay attention to the value of the expression analysis. This study took Face Reader software as an example, firstly, the facial expression recognition effectiveness in Chinese was examined by means of standard expressions database, and then the application of expression index in usability evaluation was explored using Face Reader as a tool. This research provided a scientific basis for using expression analysis method in product usability evaluation.This paper included two studies. The details were as follows:Study1 : Investigate effectiveness of Face Reader expression analysis software on the facial expression recognition in China.This part of the study used three kinds of image from USTC-NVIE standard expression database: artificial facial emotion, spontaneous facial expression, dynamic expression triggered by standard emotion stimulation. By comparing recognition rate and strength consistency between Face Reader expression analysis software and human evaluators of three kinds of expression stimulation, the study investigated effectiveness of Face Reader expression analysis software on the facial expression recognition in China.Firstly, for artificial facial expression images and spontaneous facial expression pictures, standardized select was firstly conducted. Then the recognition rate and strength consistency from Face Reader and evaluators was compared in different expression types.Secondly, for facial expression videos, standardized emotion eliciting stimulus was used to collect expression video material, then recognition rate and intensity consistency was compared from Face Reader and subjects’ self-report, Face Reader and evaluators in different expression types.Study2 : Application research of facial expression in software usability testing.This study mainly focused on two different usability of similar software(two music players) and the same software usability levels(different versions of the music player before and after optimization), compared the Face Reader expression as usability evaluation index’s effectiveness in product usability evaluation.In experiment1, typical task operation method was used to compare the usability of two different software on operation time, correct rate, subjective evaluation and expression, and then the facial expression and subjective evaluation were compared to investigate the application of expression in different software usability testing.Based on experiment1, in experiment2 the poor usability software was improved based on usability principles, then usability was compared before and after improvement on operation time, correct rate, subjective evaluation and expression, and then the facial expression and subjective evaluation were compared to investigate the application of expression in software improvement usability testing.Conclusions :(1) Face Reader software has a good ability to identify and analyze Chinese facial expression. For artificial facial expression images, 71% images can be correctly classified by Face Reader. There was significant expression intensity correlation between Face Reader and evaluators on “Sad”, “Happy”, “Surprise” and “Disgust”, “Neutral” pictures.(2) For spontaneous facial expression images, 70.6% images can be correctly classified by Face Reader, There was significant expression intensity correlation between Face Reader and evaluators on “Sad”, “Happy”, “Surprise”, “Angry” and “Disgust” pictures.(3) For facial expression videos, the correct classification rate was 32% compared to self-report. “Neutral”, “Happy”, “Sad” expressions can be identified, while “Disgust”, “Angry”, “Fear” were failed. For “Neutral”, “Happy”, “Sad” expression videos, facial expression intensity of Face Reader was significantly related with self-report; the correct classification rate was 53% compared to self-report results. “Neutral”, “Happy”, “Sad” facial expression recognition rate was higher, the “Disgust”, “Angry”, “Fear” facial expression recognition rate was lower. For “Neutral”, “Happy” and “Sad” facial expression videos, the expression intensity of Face Reader was significantly related to evaluators.(4) For two similar software products with different levels of usability, the results of this study showed that the low level usability product has longer operation time, lower correct rate, more negative expressions, lower SUS scores, and significant negative correlation between expression intensity and SUS scores on “Sad”, “Angry”. Possitive expression index can hardly reflect different levels of usability of similar software products.(5) For the same software products before and after the improvement, the results of this study showed that the low level usability product, which is the beforeimprovement one, has longer operation time, lower correct rate, more negative expressions, lower SUS scores, and significant negative correlation between expression intensity and SUS scores on “Sad”, “Angry”, “Surprise”. Possitive expression index didn’t show significant difference.
Keywords/Search Tags:Facial expression recognition, FaceReader, Usability evaluation
PDF Full Text Request
Related items