Research On Visual Enhancement Audiometric Micro-Expression Recognition In Deaf Children Based On Deep Learning

With the continuous progress and development of the times,computer technology based on deep learning appears more and more in people’s lives,and constantly solves some new problems.Deaf children,especially young children,have a series of effects on their growth due to hearing loss.At present,visual enhanced audiometry is used to judge the minimum hearing threshold,and the existence of false hearing threshold is due to the deaf child’s own factors and the lack of experience of audiometry.To solve such problems,this paper can effectively solve the problem of minimum hearing threshold for deaf children by using deep learning technology.In this study,the micro-expression recognition technology was analyzed in depth,and the bimodal state was used to identify the state of deaf children in visual reinforcement audiometry,that is,micro-expressions were used as the main modality for micro-expression recognition analysis,and eye tracking was used as an auxiliary modality for recognition,so as to judge the subject’s response to sound stimuli and assist the testers to accurately determine the hearing threshold of deaf children.It focuses on micro-expression recognition based on improved optical flow method and convolutional neural network,micro-expression recognition based on transfer learning,and the visual reinforcement audiometric assistance system developed on this basis.To solve the problem that traditional methods are susceptible to the influence of illumination and excessive computation,the improved optical flow method and convolutional neural network are used to automatically extract and classify the spatial features of feature images,and verify them on the public micro-expression datasets CASMEII.and SMIC,and the recognition rate can reach 72.78% and 74.56%.In order to solve the problem of uneven sample distribution and small sample of the dataset,and to apply the adult dataset to the visual reinforcement audiometric micro-expression recognition of deaf children,and use transfer learning for micro-expression recognition,the results show that the micro-expression recognition method based on transfer learning is 4.87% and 4.34% higher than that based on improved optical flow method and convolutional neural network,which can effectively prevent overfitting in the model training process.To solve the recognition of deaf children’s visually enhanced audiometric micro-expressions,the developed visual reinforcement audiometric assistance system for deaf children can help audiometers improve work efficiency and accurately determine the minimum hearing threshold of deaf children.Through experimental results,the efficiency of this system can be improved by 46%,and the minimum hearing threshold of deaf children can be effectively obtained.This study provides new ideas and application directions for visual reinforcement audiometry in the future,and lays a foundation for further research and application in the field of computer technology for special populations.