Visual Content Learning in a Cognitive Vision Platform for Hazard Control (CVP-HC)

This work is part of an effort for the development of a Cognitive Vision Platform for Hazard Control (CVP-HC) for applications in industrial workplaces, adaptable to a wide range of environments. The paper focuses on hazards resulted from the nonuse of personal protective equipment (PPE). Given the results of previous analysis of supervised techniques for the problem of classification of a few PPE (boots, hard hats, and gloves extracted from frames of low resolution videos), which found the Deep Learning (DL) methods as the most suitable ones to integrate our platform, the objective of this paper is to test two DL algorithms: Single Shot Detector (SSD) and Faster Region-based Convolutional Network (Faster R-CNN). The testing uses pretrained models on a second version of our PPE dataset (containing 11 classes of objects) and evaluates which of examined algorithms is more appropriate to compose our system reasoning.