Wheelchair Accessibility Evaluation with Deep Learning using a Sensing Acceleration Data-Based Spectrogram
Kazuyuki Kaneda, Masaharu Kamohara, Osamu Shiku, Toru Kobayashi · 2020
In this study, we generate spectrogram color images from wheelchair acceleration data. Using the spectrogram images, we propose evaluation models for wheelchair vibration classification using deep learning. To evaluate the stress level for the wheelchair on roads, we administered a questionnaire to wheelchair users and classified road accessibility. Using the obtained classification model, we performed experiments with a wheelchair and evaluated accessibility at points on experiment roads.