Sparse: A reservation and computer vision-based room occupancy system for Malayan Colleges Laguna's Center for Learning and Information Resources

The Center for Learning and Information Resources (CLIR) or the library in Malayan Colleges Laguna (MCL) faces some problems regarding its usage. First, students get frustrated when they go to the CLIR just to find out that there are no more available seats. Second, its operations are affected by sa...

Full description

Saved in:
Bibliographic Details
Main Authors: Lipat, Job, Rabano, Charmaine Eunice, Mamauag, Mark Anthony, Galang, Isabel, Contreras, Jennifer O.
Format: text
Published: Animo Repository 2022
Subjects:
Online Access:https://animorepository.dlsu.edu.ph/faculty_research/11624
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: De La Salle University
Description
Summary:The Center for Learning and Information Resources (CLIR) or the library in Malayan Colleges Laguna (MCL) faces some problems regarding its usage. First, students get frustrated when they go to the CLIR just to find out that there are no more available seats. Second, its operations are affected by safety protocols amidst the pandemic. To help alleviate these problems, the researchers proposed a system called Sparse. It is a computer vision-based room occupancy detection and seat reservation system. To develop the system, the researchers performed three key activities. First, the researchers trained and benchmarked three object-detection models, namely the Faster R-CNN, RetinaNet, and SSD models. These models were tested using the SCUT Head dataset. The metrics used for comparison were the models’ average processing time and root mean squared error. Second, the researchers collaborated with the visitor and librarian resource persons to identify the necessary features for the system. Lastly, the researchers surveyed 20 prospective users, including MCL students, librarians, and faculty members, to evaluate the usefulness of the identified features. From these activities, the researchers compared the Faster R-CNN (10.321 Pre-trained RMSE, 18.139 Trained RMSE, 0.117s average processing time), RetinaNet (12.660 Pre-trained RMSE, 19.026 Trained RMSE, 0.111s average processing time), and SSD (15.351 Pre-trained RMSE, 21.900 Trained RMSE, 0.018s average processing time). Among them, the Faster R-CNN was selected because it resulted in the least amount of Root Mean Square Error while having an acceptable average processing time. Then, the features for the website that were identified for visitors are the Room Occupancy Chart, Interactive Suggestions, Library Information, and Seat Reservation. For librarians, the features that were identified are the Room Occupancy Information, Report Generation, Summary of Reserved Spots, Change Effective Capacity, Library Information Management, and Account Management, wherein the last three features are only accessible to the head librarian accounts. Third, the results of the survey show that the identified features were useful to users. Additionally, the survey has also gathered suggestions that could be implemented in the future.