2018/11/08 by Benjamin Johnston, Johnston, Benjamin, Philip de Chazal +1
Medicine · Computer Science · #Nasal Surgery and Airway Studies #Sinusitis and nasal conditions #Face recognition and analysis
paper · pdf · doi:10.48550/arxiv.1811.03773
We present a fully automated system for sizing nasal Positive Airway Pressure\n(PAP) masks. The system is comprised of a mix of HOG object detectors as well\nas multiple convolutional neural network stages for facial landmark detection.\nThe models were trained using samples from the publicly available PUT and MUCT\ndatasets while transfer learning was also employed to improve the performance\nof the models on facial photographs of actual PAP mask users. The fully\nautomated system demonstrated an overall accuracy of 64.71% in correctly\nselecting the appropriate mask size and 86.1% accuracy sizing within 1 mask\nsize.\n