Firefighting and rescue live victims operations are inherently dangerous, but the imminent danger of release of a hazardous substance creates an additional risk. Thus, identification of hazardous materials during robot assisted search and rescue missions can help e.g. firefighters or rescue teams to improve such rescue operations. The paper deals with the development of such a robotic machine vision system for hazmat label recognition. Classical computer vision methods but also state-of-the-art deep learning based detection algorithms were implemented and evaluated. Special focus was put on the robustness of detection and recognition with limited hardware resources and the influence of background image structures and light conditions.