Table 8.

A summary of AI-Big data analytics models used for occupancy detection

Ref.	AI model	Detection basis	Building nature	Year	Description	Evaluation metrics
						ACC	MAE	MSE	Others
Huang and Hao (2020)	CNN	Surveillance cameras	Office	2020	Detect the number and location of occupants	$✓$			Relative error
Acquaah et al. (2020)	CNN, SVM	Thermal cameras	–	2020	Estimate the number of people present based on thermal images	$✓$
Tien et al. (2021)	CNN	Vision cameras	Office	2021	Predict equipment use and occupancy count & activity in real time	$✓$			Precision, Recall, F1
Zhao et al. (2018)	SVR, RNN	Temperature sensor	Office	2018	Detect the number of occupants based on indoor thermal properties				Error rate
Elkhoukhi et al. (2020)	LDA, VHT	Indoor sensors	Office	2020	Predict the status of occupants’ presence	$✓$
Fatema and Malik (2021)	NN	Indoor sensors	Office	2021	Predict occupancy condition in an office room	$✓$			TPR, FPR, Precision, Recall, F1, MCC
Wu and Wang	SVM, kNN, DT RF, NN	Infrared sensor	–	2021	Provide accurate predictions of the occupancy status based on motion detectors	$✓$
Huchuk et al. (2019)	LR, HMM, MM, RF, RNN	Thermostat data	Residential	2019	Forecast the occupancy information	$✓$
Razavi et al. (2019)	kNN, SVM, NN, RF	Energy meter	Residential	2019	Estimate and predict occupancy information	$✓$			Precision, AUROC
Feng et al. (2020)	CNN, BiLSTM	Smart meters	Residential	2020	Predict real-time occupancy status based on data of electrical signals	$✓$			Precision, Recall, F1, TNR F1, Training time
Pešić et al. (2019)	LSTM	Bluetooth and WiFi devices	Residential	2019	Predict, forecast, and analyze occupancy information using wireless networks data	$✓$			RMSE and Edit Distance on Real Signals