Verification of Medication Dispensing Using the Attentive Computer Vision Approach

This paper is dedicated to the development of an assistive computer vision-based system for medication verification before dispensing it to patients, e.g., in long-term healthcare facilities. The algorithmic basis of the system is the attentive vision approach to robust and fast object detection in images. It consists of time-efficient image analysis by a multiscale visual attention operator to detect feature-point areas located inside the pill regions. The medication identification algorithm is based on the matching of pill descriptor vectors with the reference ones, which are obtained by the attentive machinelearning algorithm during the training phase. A new set of rotation-invariant image descriptors is proposed, which fully discriminates between different medication pills.