This paper presents the analysis of water supply behavior within an educational campus, serving as a use-case to demonstrate the broader applicability of an innovative IoT-based framework integrated with deep learning techniques. By retrofitting analog water meters with IoT devices, the study captures images of meter dials, which are then locally processed using a deep learning-based digit detection algorithm. This process converts the images into digits and transmits the data to the cloud for real-time analysis, thereby enhancing the accuracy and reliability of water usage data. Focusing on two key regions within the campus—student hostels and faculty/staff quarters—the analysis thoroughly examines the impact of water supply patterns on both a monthly and weekly basis. It reveals how the distinct characteristics of each month, such as holidays, exams, and class schedules, significantly influence water consumption in these areas. The study particularly highlights the variations in water usage in student hostels, driven by the academic calendar and student lifestyle, in contrast to the more stable water demand observed in faculty/staff quarters. The integration of the data refinement algorithm uncovers the underlying consumption patterns within these campus residence. The findings from this detailed investigation are instrumental in understanding the water distribution patterns, particularly within Integrated Water Systems (IWS), and set a precedent for the potential scalability and adaptability of the framework. This study not only sheds light on the specific water management needs of an educational campus but also suggests that the successful application of this system in such a dynamic and varied setting indicates its potential for broader application, thereby contributing to more informed decision-making and promoting sustainable water management practices in various contexts.