A Recognition Approach Of Design Fixation Based On Body Language

Design fixation is a phenomenon that designer keeps the blind adherence to a set of concept or idea.It leads to that some elements or features are used in different design repeatedly.Nowadays innovation design is paid more and more attention while design fixation is one of the main factors that hinder innovation.Hence,researches on design fixation are highly valued.A research reported that the negative effects of design fixation will be reduced if the designer gets inspiration when the fixation appears.However,the approach of recognizing design fixation still needs to be studied.The artificial intelligence has matured and becomes more reliable in recent years,especially in the field of computer vision.It provides a new method to the study of design fixation.This paper focused on the behavior of the designers in the stage of the conceptual design.The relationship between design fixation and body language was studied by experiments.A recognition system based on machine learning was developed and its availability was proved by a design case.Firstly,we conducted an experiment of body language on 12 subjects.Their behavior was recorded by a camera and their design sketches were collected.An echogram was established by counting the frequency of each behavior.Secondly,the fixation degree of each sketch was evaluated by 6 experts.The influence of different fixation degree on behavior was analyzed by ANOVA and the significant difference of behavior on fixation was obtained.Thirdly,a model of deep convolutional neural network based on VGG16 was built and achieved 88.68% accuracy.It could recognize 3 kinds of behavior which are related to design fixation.Another model of classifier based on SVM was also built and achieved 72.31% accuracy.It could judge whether the designer is in fixation or not throughout the statistics of behavior.Finally,a recognition system based on these two models was developed and its usability was proved by a design case.