"Is Bukang-i Here Now" Developed in a Day… Pukyong National University Student Creates AI Reporting Service

Citizens Submit Sightings and Photos Directly

AI Conducts First Screening to Verify On-Site Photos

20,000 Visits Recorded in Two Weeks Since Launch

A new "hot" service has emerged for citizens searching for "Bukang-i," the shark spotted at Busan North Port. This web service allows citizens to report shark sightings themselves, and an AI reviews the photos and displays the information in real time.

The "Is Bukang-i Currently Available" service, developed by Inoa—a master's degree student affiliated with the ICT Core Talent Training (ICAN) Program for Integrated Bachelor's and Master's Degrees at Pukyong National University—has drawn notable attention, recording approximately 20,000 cumulative visits in just two weeks since its launch.

Inoa, a researcher in Professor Choi Sungcheol’s TEAMLAB at the Department of System Management and Safety Engineering at Pukyong National University, created the service in a single day after a shark appeared in the canal of the North Port Waterfront Park and crowds gathered. The service began operation on September 22, 2026, following its development after September 18, when the shark first appeared.

The method of use is simple: Citizens at the site can click either "Visible Now" or "Not Visible Now" and upload a photo, after which the sighting information is shown on a map of the park. Users who submit a photo can also receive a verification card.

Initially, Inoa manually searched for and transferred sighting posts from the Threads social media platform onto the map. However, as the number of reports increased, it became difficult for a single operator to review and update the situation one by one.

The service was later modified so that citizens themselves could directly enter information at the site. The "Visible Now" and "Not Visible Now" buttons could only be used within the park’s vicinity, and photo submissions were subject to an AI's initial verification.

Is Bukang currently available on the service screen?

The AI simultaneously determines whether an actual shark is present in the submitted photo and checks whether the photo was taken on-site. The system is designed to filter out images not shot at the location, such as screenshots from TV broadcasts.

Reports with ambiguous results are reviewed and approved by a human operator. By implementing an "AI first judgment – human final approval" system, reliability has been enhanced, rather than relying solely on AI to make all decisions.

The identification module was changed once during service operation. Initially, a commercial AI tool was used, but this resulted in high costs and intermittent errors. The architecture was then switched to an open-source JEV-based inference structure, which maintained accuracy while boosting processing speed by more than five times.

Inoa applied experience gained through research and coursework conducted in the ICAN program at Pukyong National University to develop this service. The ICAN program aims to cultivate core ICT talent by connecting undergraduate and master's degree coursework, and it offers a "Digital Twin AI and Metaverse XR Research Education Program." In this program, companies present real-world problems and data, and students solve them through projects.

Through this process, Inoa worked on practical tasks such as automating research for news classification and recruitment document evaluation. Through these projects, Inoa learned to design systems that filter out AI errors and record the basis for judgments. This experience influenced the AI first judgment and human final approval structure adopted in this service.

Inoa explained, "At first, I manually searched Threads and transferred reports, but there were clear limitations. Now, citizens can submit reports with just one click, and AI automatically reviews the photos, allowing required information to be uploaded from the field much more quickly."

Inoa also added, "What helped most when urgently developing the service was what I learned in the lab about making evidence-based decisions, rather than simply forming plausible answers."

Professor Choi Sungcheol noted, "It is meaningful that a student transferred the skills of problem definition and evaluation design gained in the program into a service that is now used by citizens."

He added, "The ICAN program will continue to support students so they can solve real-world problems firsthand and gain experience presenting results to society."

Inoa, Master's Student at Pukyong National University.

This content was produced with the assistance of AI translation services.