Table 2.

Facial expression/emotion.

Reference	Focus	N participants	Age	Input data/device used	Method used	Dataset
Leo et al.47	Facial expression for quantitative assessment	17 ASD, 10 TD	6–13 years	Image sequences	Deep learning	Own dataset
Kalantarian et al.36	Facial emotion for mobile games	8 ASD	6–12 years	Mobile phone	Ensemble classification (AWS + Sighthound + Azure)	Own dataset
Kalantarian et al.37	Facial expression for quantitative assessment	8 ASD, 5 TD	ASD: 8.5 ± 1.85 TD: 4.4 ± 0.54 (in years)	Video, mobile phone	Histogram of Oriented Gradients (HOG) + SVM	Own dataset
Han et al.38	Emotional expression recognition	25 ASD		Camera	Deep learning, CNN	128,129
Tang et al.39	Automatic smile detection	11 ASD, 23 TD	6–24 months	Video, two wireless cameras	Deep learning, CNN	GENKI-4K, CelebA132, RCLA&NBH Smile
Daniels et al.40	Emotion recognition for assistive technology	23 ASD, 20 TD	6–17 years	Google Glass		n/a
Jazouli et al.41	Emotion recognition for assistive technology	10 ASD		3D image, Microsoft Kinect		Own dataset
Washington et al.42	Emotion recognition for assistive technology	14 ASD	9.57 months [3.37. 4–15]	Video/Google Glass and mobile phone	Machine learning, Histogram of Gradients (HOG) + SVM	128,139–143
Voss et al.43	Emotion recognition for assistive technology	20 ASD, 20 TD		Video/Google Glass and mobile phone	Machine learning, Histogram of Gradients (HOG) + SVM	n/a
Vahabzadeh et al.44	Emotion recognition for assistive technology	8 ASD	11.7–20.5 years	Video, Google Glass		n/a
Leo et al.45	Emotion recognition for behaviour monitoring	3 ASD		Video, Robokind R25 Robot		128
Pan et al.46	Facial emotion for behaviour analysis	2 ASD		Video, NAO robot		Own dataset
Coco et al.48	Facial expression analysis for diagnosis	5 ASD, 5 TD	65.38 months [15.86, 48–65 months]	Video, webcam	Deep learning, Histogram of Oriented Gradients (HOG) feature combined with a linear classifier, CNN	DISFA [24], SEMAINE [26] and BP4D [34] datasets.
Leo et al.49	Facial expression for quantitative assessment	17 ASD	6–13 years	Image sequences	Deep learning	Own dataset
Samad et al.50	3D facial imaging for physiology-based impairment detection	8 ASD, 8 TD	7–20 years	3D images, high resolution 3D facial imaging sensor, 3dMD		n/a
Leo et al.51	Facial expression recognition for assistive technology	1 ASD, 1 TD		Video	Deep learning, Facial Action Coding System (FACS)	Own dataset
Guha et al.52	Facial expression for quantitative assessment	20 ASD, 19 TD	9–14 years	Motion capture data, 6 infra-red motion-capture cameras	Deep learning, Facial Action Coding System (FACS)	Own dataset
Ahmed and Goodwin53	Facial expression for predicting engagement and learning performance	7 ASD	8–19 years	Video, camera	Computer Expression Recognition Toolbox	Own dataset
Harrold et al.54	Facial expression for assistive technology	2 ASD, 4 TD	8–10 years	Video, Apple iPad		n/a
Harrold et al.55	Facial expression for assistive technology	2 ASD, 4 TD	8–10 years	Video, Apple iPad		n/a
White et al.56	Facial emotion expression and recognition	20 ASD, 20 TD	9–12 years	3D data, Microsoft Kinect		n/a
Garcia-Garcia et al.57	Facial expression for learning emotional intelligence	3 ASD	8–10 years	Video, mobile phone	Affectiva SDK	n/a
Jain et al.58	Facial expression recognition for assistive technology	6 ASD	5–12 years	Video, webcam		128
Li et al.59	Facial attributes for ASD classification	49 ASD, 39 TD		Video, Apple iPad	Deep learning, CNN	Training: AffectNet133 and EmotioNet134 Evaluation: Own dataset
Shukla et al.60	Facial image analysis for diagnosis	91 ASD, 1035 NDD, 1126 TD		Image, camera	Deep learning, CNN	Own dataset