Open AccessComparison of the YOLOv3 and Mask R-CNN architectures’ efficiency in the smart refrigerator’s computer vision
Author(s) -
Mikhail Dorrer,
A. E. Tolmacheva
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1679/4/042022
Subject(s) - computer science , set (abstract data type) , refrigerator car , task (project management) , artificial intelligence , table (database) , computer vision , test set , architecture , selection (genetic algorithm) , engineering , data mining , mechanical engineering , art , systems engineering , visual arts , programming language
The article deals with the computer vision system of the smart refrigerator “Robimarket”. The equipment of the working area of the refrigerator, the selection of a set of chambers, the collection of a training sample for the computer vision system are described. The choice of the artificial intelligence architecture of the computer vision system was made by comparative testing of the YOLOv3 and Mask R-CNN architectures. The comparison was made on one hardware platform, one training set and a set of test cases. As a result, a comparison table was created for the speed and quality values of each model. As a result, the Mask-RCNN architecture was chosen, which showed a significantly higher detection accuracy in the video stream with acceptable performance for this task.