Ryumina, II; Kukhartseva, MV; Narogan, MV; Borovikov, PI; Lagutin, VV; Whiteley, I; (2021) The use of computer vision technologies for an objective assessment of indicators of concentration of attention in newborns and infants with visual stimulation for the purpose of developmental care. Neonatology: News, Opinions, Training , 9 (1) pp. 30-41. 10.33029/2308-2402-2021-9-1-30-41.

Preview

Text Whiteley_The use of computer vision technologies for an objective assessment of indicators of concentration of attention in newborns and infants with visual stimulation for the purpose of developmental care_AAM.pdf - Accepted Version Download (800kB) | Preview

Abstract

Assessment of visual function in the first few days after birth is mainly limited to the study of eye movements, the ability to fix a gaze and follow an object. In order to determine how the child's gaze moves when looking at an object, how long it is fixed on the object, techniques are needed to record the movements of the eyeballs and determine the trajectory of the gaze. THE AIM OF THE study is to develop a method based on machine learning technology and computer vision for the automated analysis of eye movement and fixation of the gaze of a newborn and an infant during visual stimulation. MATERIAL AND METHODS: The proposed method includes video filming of newborns and children of the first year of life in hospital and outpatient. Video recordings from mobile phone cameras with a length of 15 s to 3 min were used as the initial data. Of the 150 videos of newborns, 73 videos were selected, of which 61 recordings of sufficient quality were selected, in which the child was recognized in at least 30% of frames. For each recording, a track was recorded of the state of the newborn during visual stimulation. Facial recognition was implemented using a widely used pre-trained model based on machine learning and ultra-precise networks. The algorithm for the study of eye movement includes searching for a face, determining the position of the head by the location of the eyes, nose, lips, searching for the eye zone, searching for the pupil, determining the relative and absolute direction of gaze. RESULTS AND DISCUSSION: A neural network was developed and trained for recognizing facial images of newborns and babies and for locating the eyes on a child's face. The method made it possible to obtain data on the direction of the child's gaze in real time using the camera of an ordinary smartphone or a simple web camera. Depending on the size of the displayed image and the distance to it, the system calculates the total time of concentration on the image, and also detects moments when the child is not interested in the image. CONCLUSION: The proposed method can be used to analyze the effectiveness of early visual stimulation in children, in the context of long-term effects on psychomotor and cognitive development, as well as to objectify data from various programs for early assessment of visual function in newborns and infants.

Download activity - last month

Export as CSVJSONXML

Download activity - last 12 months

Export as XMLJSONCSV

Downloads by country - last 12 months

Export as CSVJSONXML

Archive Staff Only

View Item