An Integrated Multi-Stage Computer Vision System for Real-Time Detection of Potential Child Labor in Hazardous Industrial Environments

Child labor in dangerous workplaces-such as construction zones, industrial sites, and workshops-remains a critical humanitarian and social challenge, threatening children's health, safety, and future.Addressing this requires proactive, technology-driven solutions that can operate effectively in real-world conditions.This paper presents a multi-stage, real-time computer vision system for early detection and prevention of child labor in hazardous environments.The system starts with a ResNet50-based scene recognition model trained on 6,000 labeled images, classifying industrial settings with 96.2% accuracy.A YOLO-powered object detection module, trained on 6,000 annotated frames, identifies individuals in video feeds, prioritizing the largest bounding box for further checks.The same YOLO framework detects Personal Protective Equipment (PPE) such as helmets, vests, and gloves, achieving a mean average precision (mAP) of 94.7% for compliance verification.The system's face detection module identifies facial regions within the input stream with an accuracy of 92.06%, evaluated on a dataset comprising approximately 90,000 images.Once a face is detected, the corresponding region is extracted and passed to a demographic estimation model.This model, developed using the FairFace architecture and trained on 96,509 images, predicts the individual's age with an accuracy of 79.20%.If the predicted age indicates that the person may be under the legal threshold, the system promptly triggers a real-time alert, ensuring rapid response and intervention.Designed for integration into existing surveillance infrastructures, the system provides immediate alerts to authorities, enabling rapid intervention.Testing across varied industrial settings confirms robustness, speed, and adaptability.This scalable and ethical tool supports NGOs, government agencies, and industries committed to eliminating child labor, promoting safer and more humane workplaces.