“…x Dong et al [16] x RH, temperature, light, CO2 Polese [75] Image processing occupancy sensor Labeodan et al [39,76] Chair sensors Jazizadeh and Becerik-Gerber [77] Light intensity sensors x Liao et al [78] Beam-break sensors Meyn et al [79] Camera, CO2 Tiller et al [48] x Wired SN Dedesko et al [81] Beam-break sensors, CO2 Erickson et al [82] Wireless Camera Erickson et al [17] Wireless Camera Harle et al [83] Ultrasonic 3D tracking Benezeth et al [84] Camera Hailemariam et al [29] RH, temperature, light, AC current Sensor array Attar et al [85] RH, temperature, light, AC current Sensor array Han et al [86] RH, temperature Han et al [87] CO2, PIR, RH Pandharipande and Caicedo [88] Ultrasound array sensor Agarwal et al [21] PIR + door sensor x Weekly et al [89] Particulate matter sensor Yavari et al [90] Doppler Radar Arora et al [54] Illuminance, RH, PIR, power consumption, door contact, CO2 Gunay et al [45] x PIR, door contact, CO2 Philipose et al [91] x Ramoser et al [92] x Brackney et al [93] Image processing occupancy sensor Milenkovic and Amft [94] PIR, ultrasound range finders, and plug-in equipment load Layered hidden Markov models were trained with 30 h worth of data for presence detection and with 50 h worth of data for people count detection. The models were shown to detect presence accurately.…”