Table 2.

Literature review related to algorithms

Ref. No.	Paper Proposal	Hardware/Software	Datasets	Applied Algorithms	Accuracy
[54]	This paper proposes an electronic system equipped with wearable sensors for gesture recognition to interpret abnormal activities of the children to their parents through a machine learning algorithm.	Flex Sensor, Arduino, mpu6050, Bluetooth	Alphabets, 20 iterations	KNN	95%
[55]	This work provides ICT solutions for autistic children by examining a person’s voice, body language, and facial expressions to monitor their behavior while performing gestures.	Accelerometer, PC, RGB camera	40 gestures, 10 persons, 10 repetitions	Parallel HMM	99%
[56]	This article presents the links between known attention processes and descriptive indicators, emotional and traditional gestures, and nonverbal gestures between ASDs in attention processes and gestures.	Flex sensor, MPU 6050, contact sensor, Arduino, Bluetooth, PC	1300 words	HMM	80%
[57]	This paper has proposed a system in which the recognition process uses a wearable glass and allows these children to interpret their gestures easily.	RGB camera, PC	26 postures, 28,000 images	HAAR cascade algorithm	94.5%
[58]	This paper presented the idea to analyze the development of conventional gestures in different types of children, such as typical ASD children.	RGB camera, PC	English and Arabic alphabets	BLOB, MRB, least difference, GMM	89.5%, 80%
[59]	This paper presented the IoT-based system called” Wear Sense” to detect the atypical and unusual movements and behaviors in children who have ASD.	Capacitive touch sensor, R-pi, Python	36 gestures, 30 trials each	Binary detection system	86%