Object Classification for Child Behavior Observation in the Context of Autism Diagnostics Using a Deep Learning-based Approach (CROSBI ID 419442)
Ocjenski rad | diplomski rad
Podaci o odgovornosti
Presečan, Mihael
Kovačić, Zdenko
Petric, Frano
engleski
Object Classification for Child Behavior Observation in the Context of Autism Diagnostics Using a Deep Learning-based Approach
This research sets the ground for the new task in ADORE protocol - the free play observation. The core problem to solve is to develop accurate, robust and fast object detector. This research have proposed Deep Learning approach for object detection and classification for the child behavior observation task. The task is used in the context of autism diagnostics, which uses the ADOS protocol with it’s standardized toys. Child behavior has been determined by it’s playing with the toys, and in order to detect the toys and the child, a new dataset has been developed. Also, this research had developed algorithms and mechanism to determine child’s attention based on the toys that child is playing with. This thesis disscuses the challenges encountered in the research and their solutions, and as well sets the work for continuing the development of robust and accurate attention analyzer.
deep learning, object detection, object classification, Faster R-CNN, ADORE, ADOS, ADOSet, free-play, autism
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o izdanju
54
17.07.2017.
obranjeno
Podaci o ustanovi koja je dodijelila akademski stupanj
Fakultet elektrotehnike i računarstva
Zagreb