Labor exploitation investigation using statistical and multiple object tracking assessment methods

P. Karthikeyan,Chih Chun Chang,Pao-Ann Hsiung
DOI: https://doi.org/10.1007/s11042-023-16094-2
IF: 2.577
2023-07-12
Multimedia Tools and Applications
Abstract:Labor exploitation in Taiwan's Distant Water Fishing (DWF) industry has been a significant issue for many years. Migrant fishermen from Southeast Asian countries, including Indonesia and the Philippines, have reported being subjected to poor working conditions, long work hours, and low pay. Our research aims to identify and address labor exploitation in Taiwan's DWF vessels through the development of a three-module system. The first module is a mobile application interface used to collect data from fishermen and captains. The second module is responsible for collecting data from an offline SQLite database and Closed-circuit television (CCTV) footage from DWF vessels. The third module employs fisherman detection and tracking models to analyze working hours. We have developed two models to analyze labor exploitation in DWF vessels: a statistical assessment method and a Multiple Object Tracking (MOT) assessment methods. The statistical assessment method provides a quick response, while the MOT assessment method tracks all fishermen in CCTV footage and computes their 24-h work time, which is then compared with the mobile application data to identify instances of labor exploitation. We applied statistical assessment methods to analyze CCTV footage from January 25th, 2022, to January 31st, 2022, and MOT assessment methods were applied to CCTV footage from February 7th, 2022, to February 13th, 2022. Our analysis indicates that there were no instances of labor exploitation during this time frame.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering
What problem does this paper attempt to address?