Occupancy Estimation Using a Low-Pixel Count Thermal Imager

An occupancy estimation sensor system based on low-pixel count sensor arrays is proposed. We evaluate a system comprising a 4 × 16 thermal detector array. Machine learning classifiers are used to interpret the raw data from the detector array for deducing the number of occupants in the sensor's field of view. We observe that nominal classifiers provide more robust classification than numerical classifiers. Furthermore, entropy-based classification is applied for the first time for occupancy estimation and found to produce the lowest root mean-squared errors and highest correlation coefficients compared with previously trialled classifiers.