We will have a special hybrid seminar on December 17 in which Mr. Decio Yoshimoto (PhD Candidate at Instituto Tecnologico de Aeronautica -Brazil) will give a talk on “Airport Service Quality Assessment through Machine Learning-Enhanced Computer Vision Metrics”

If you are interested, please check below.

[The 25th UTokyoIPL-CUTI seminar]

1) Time and day: 10:30-12:00 (Japan Standard Time), December 17th, 2025 (Wed)

2) Place: Seminar room of International Project Lab, Department of Civil Engineering, UTokyo (3rd floor of Engineering Building No.11, Hongo Campus, UTokyo) and online

https://u-tokyo-ac-jp.zoom.us/j/87434147471?pwd=WZHTxbTQOCX1fsbnRUQWxWHA0jD7rb.1

Meeting ID: 874 3414 7471

Passcode: 477050

3) Presentation

– Presenter: Decio Yoshimoto (PhD Candidate at Instituto Tecnologico de Aeronautica -Brazil)

– Title: Airport Service Quality Assessment through Machine Learning-Enhanced Computer Vision Metrics

– Abstract: This study introduces qLoS, a quantitative Level of Service metric for airport check-in areas enabling automated, continuous assessment through computer vision. qLoS integrates passenger-weighted parameters for queueing time, path/distance, queue density, and baggage dynamics within IATA's standardized framework, implemented through a computer vision system combining YOLO11 detection with DeepSORT tracking and homography transformation with multi-pole calibration for position estimation in severely occluded environments.

The metric addresses critical limitations in airport LoS research where traditional manual observation methods are constrained to small sample sizes and limited observation periods, while simulation models depend on predetermined variables and theoretical passenger arrival patterns rather than empirical behavior. This data scarcity—with existing studies analyzing dozens to hundreds of passengers over limited periods—prevents comprehensive understanding of real passenger journeys, infrastructure constraints, and stochastic behavior patterns. The system was validated through controlled laboratory experiments with marked floors and Blender simulation environments, computing comprehensive per-passenger metrics from continuous video data.

This research advances airport operations by replacing subjective surveys and simulation- based analysis with objective, extensive datasets enabling data-driven resource allocation and capacity planning. The multi-phase methodology establishes foundation for machine learning integration and predictive qLoS modeling with operational airport data.

4) Short bio of presenter

Decio Yoshimoto is a PhD candidate in Aeronautical Engineering at the Instituto Tecnológico de Aeronáutica (ITA) coupled with a Massachusetts Institute of Technology’s (MIT) Advanced Study Program. His research bridges airport operations and economics: he develops homography-based, uncertainty-aware measurement from overhead video to recover passenger trajectories, queue density, and queue Level-of-Service (qLoS), and evaluates the economic impact of airport investments and efficiency interventions on throughput, service quality, and cost. He teaches Advanced Finance Strategies and Valuation in a leading MBA program and at ITA, and previously taught Corporate Finance at GM University (Detroit). Yoshimoto’s international industry experience spans the Americas, Africa, Asia, and the Middle East with organizations such as General Motors and Brookfield Asset Management and multilaterals including IFC, EDC, and the IDB; he also led the first successful small-hydropower carbon-emission certificate issuance and shared insights as a guest of the United Nations and IFC at COP11. He holds a BS (Instituto Mauá), an MBA (The Ohio State University), and an executive program certificate from Northwestern University’s Kellogg School of Management.

5) Charge: free

6) Language: English

7) Participation: Please contact Ms Tomoko Samukawa (samukawa@ip.civil.t.u-tokyo.ac.jp) for joining this seminar, but you can join the seminar even without pre-registration.