This paper presents a passive UHF RFID (Ultra High-Frequency Radio Identification) tag array-based occupancy detection of smart tourism using the AI (Artificial Intelligence) method of (RF) Random Forest. To illustrate the randomness of occupancy in different scenarios, this paper compromises UHF RFID tag array sensing and AI computing into an open platform for occupancy detection in different scenarios. Our proposed system aims at predicting population density classification from 0% to 75%. To demonstrate the feasibility of our approach, we conducted a small-scale experiment using a carpet embedded with nine passive UHF RFID tags on average. We deployed two different tag placement patterns, Circular and Square, to assess their impact on classification accuracy. During the experiment, one person walked on the carpet, covering approximately 25% of its area. We assigned each participant to a corresponding occupancy class based on the carpet percentage covered in 4 classes, corresponding to 0, 1, 2, and 3 people walking on the carpet, e.g., class 1 represents 0 individuals, class 2 represents one person, and so on one at a time. We could accurately classify each participant into the appropriate occupancy class by analyzing the data collected from the RFID tags. The employed AI method of RF achieves high classification accuracy of 91.82%, which is much higher than other common classifiers of SVM and BP Neural Network.