The ACM International Conference on Advances in Geographic Information Systems 2021 (ACM SIGSPATIAL 2021) will be held on November 2-5, 2021 at online conference. The following presentations will be delivered.
1. CityOutlook: Early Crowd Dynamics Forecast towards Irregular Events Detection with Synthetically Unbiased Regression.
When a large-scale event is held in a city, the technique of “early crowd dynamics forecasting,” which predicts the crowd density at the venue at an early stage, is very important as a means of crowd control or COVID-19 countermeasures.
Conventional methods for this topic have been proposed that take into account external factors such as holidays and weather, as well as users’ future activity schedules [Konishi+ UbiComp “16; Anno+ UbiComp’20].
However, events that induce congestion are rare and it is difficult for conventional methods to learn this rare pattern, making it challenging to quantitatively estimate the number of visitors when congestion occurs.
In this paper, we focus on importance weighting to suppress the increase of model bias which was the cause of the difficulty. The proposed method, CityOutlook, can effectively learn rare patterns while theoretically guaranteeing the reduction of the model bias by using the importance weighting defined by the probability density ratio of crowded patterns to uncrowded patterns.
To make importance weighting applicable to the crowd dynamics regression problem, we also proposed a framework for assigning labels to the data for crowded and uncrowded patterns, taking into account external factors and user’s schedules.
Through empirical experiments using real data, we show that the proposed method can both quantitatively estimate the number of visitors under congested conditions and stably predict the crowd density under un-congested conditions.
2. AI-BPO: Adaptive incremental BLE beacon placement optimization for crowd density monitoring applications
With the pandemic of COVID-19, indoor crowd density monitoring has become significant to the public health-care. The University of Tokyo released an indoor density monitoring system using BLE beacons to monitor the indoor crowd density. In order to obtain accurate indoor crowd density monitoring results, we carefully designed beacon placement to cover the spatial area with consideration of the environmental factors.
Beacon placement optimization has been studied for years. Some researchers proposed batch simulation-based sensor placement optimization methods to obtain the optimal placement by simulating radio wave propagation of beacons. However, the only simulation cannot reflect the actual radio map in the target environment. Some other research proposed optimization methods by selecting beacons from large distributed beacons. This approach can provide the beacon placement considering actual radio propagation. However, it demands an ideal initial beacon placement of adequate beacons and dense data measurement.
In this research, we propose the method of adaptive incremental beacon placement optimization (AI-BPO), which incrementally determines the new beacon placement location after gathering RSSI data in the environment with less labor cost. We experimented with the university buildings, and the experimental results show the effectiveness of our proposed method.
— Presentation information —
Poster/Demo Session 1B (Wed., Nov. 3, 2021, 12:30 PM – 2:00 PM JST (Tue., Nov. 2, 2021 08:30 PM – 10:00 PM PDT))
Soto Anno and Kota Tsubouchi and Masamichi Shimosaka.
CityOutlook: Early Crowd Dynamics Forecast towards Irregular Events Detection with Synthetically Unbiased Regression.
SIGSPATIAL’21: Proceedings of the 29th International Conference on Advances in Geographic Information Systems, November 2021, Beijing, China, November 2021.
Presentation on YouTube: https://www.youtube.com/watch?v=-yrDJXDk_1E
Yang Zhen, Masato Sugasaki, Yoshihiro Kawahara, Kota Tsubouchi, Matthew Ishige, Masamichi Shimosaka
AI-BPO: Adaptive incremental BLE beacon placement optimization for crowd density monitoring applications
SIGSPATIAL’21: Proceedings of the 29th International Conference on Advances in Geographic Information Systems, November 2021, Beijing, China, November 2021.
Presentation on YouTube: https://www.youtube.com/watch?v=5qAT7qljAfc