Calculation of Museum Collections Popularity Using Detection and Tracking of Visitors From Surveillance Cameras in Low Light Conditions

The Museum has attempted to measure the level of visitor interest in the collections on display; this is important for curatorial decisions and improving the visitor experience. Traditional methods such as surveys and observations, which tend to be intrusive and inaccurate, are being replaced by technological advances. Previous research has explored visitor tracking and engagement using a variety of technologies, including sensors, Bluetooth signals, RFID tags, QR Barcode, and Single Board PC Sensors. However, these methods still require tools or other devices to work. This research introduces a new approach to measuring museum collections' popularity by detecting and tracking visitor interest using surveillance cameras in low-lighting museum environments. We developed a new model with the YOLOv8 algorithm using a custom image dataset from the museum environment to detect and track visitors from surveillance camera videos enhanced using the CLAHE (Contrast Limited Adaptive Histogram Equalization) method in regions of interest (ROI) around the museum collection. The results show that the accuracy of the custom model is 86.5% which is effective in detecting visitors, and with the enhancement of video quality, it is successful in detecting and tracking visitors by gaining 3% more 'passing visitors' and 22% more in 'interested visitor' in museum collection popularity.