eduzhai > Applied Sciences > Engineering >

DeepHAZMAT Hazardous Materials Sign Detection and Segmentation with Restricted Computational Resources

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 14 pages

Abstract: One of the most challenging and non-trivial tasks in robot-based rescueoperations is the Hazardous Materials or HAZMATs sign detection in theoperation field, to prevent further unexpected disasters. Each Hazmat sign hasa specific meaning that the rescue robot should detect and interpret it to takea safe action, accordingly. Accurate Hazmat detection and real-time processingare the two most important factors in such robotics applications. Furthermore,we also have to cope with some secondary challenges such as image distortionand restricted CPU and computational resources which are embedded in a rescuerobot. In this paper, we propose a CNN-Based pipeline called DeepHAZMAT fordetecting and segmenting Hazmats in four steps; 1) optimising the number ofinput images that are fed into the CNN network, 2) using the YOLOv3-tinystructure to collect the required visual information from the hazardous areas,3) Hazmat sign segmentation and separation from the background using GrabCuttechnique, and 4) post-processing the result with morphological operators andconvex hull algorithm. In spite of the utilisation of a very limited memory andCPU resources, the experimental results show the proposed method hassuccessfully maintained a better performance in terms of detection-speed anddetection-accuracy, compared with the state-of-the-art methods.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×