Using machine learning and blockchain for trusted detection and monitoring of excessive working hours in factories

The Organisation for Economic Co-operation and Development (OECD) Due Diligence Guidance emphasizes the importance of managing working hours to protect the rights of workers. Excessive hours pose serious health risks, which highlights the need for robust detection and reporting systems. However, many of today's systems, methods, and technologies used for managing labor hours lack traceability, auditability, accountability, and trust. Additionally, they are centralized and manual or paper-based, which makes them vulnerable to manipulation as they are controlled by a limited number of entities. In this paper, we present a machine learning and blockchain-based solution to automate the detection of excessive working hours in a manner that is decentralized, as part of an antitrust coalition, with regulated transparency, traceability, auditability, and trustworthiness. We develop smart contracts to automate compliance reporting and manage large datasets off-chain through decentralized storage. The proposed system achieves a detection accuracy of 96.6% and a precision of 92%. We conduct a comprehensive evaluation of the proposed solution, including cost analysis, security assessment, and performance evaluation of the worker detection component. By comparing our solution to existing safety monitoring systems, we demonstrate its superior automation, traceability, and trustworthiness. The proposed solution not only enhances worker safety and compliance with OECD guidelines but also contributes to sustainability in industrial environments.