Home

Menu

Loading wiki pages...

View
Wiki Version:
<p><strong>Hazmat Label Test Image Dataset (HLTID)</strong></p> <p><img alt="Example input images with corresponding ground truth maps" src="http://getwww.uni-paderborn.de/~jeti/hazmat-dataset.jpg"> <em>Example input images (top row) with corresponding ground truth maps (bottom row).</em></p> <p>Hazmat labels indicate dangerous substances and are of high relevance for rescue robots and other security and safety systems. Their detection and identification is therefore and important task in the field of computer vision. </p> <p>The image dataset we provide here contains 600 high-resolution photographs of hazmat labels placed on different backgrounds and corresponding high-accuracy ground truth maps (see figure).</p> <p>The 600 images result from the following factors:</p> <ul> <li>3 different backgrounds: Woodchip wallpaper, OSB wood, and bricks wall.</li> <li>8 different hazamat types: combustible, dangerous when wet, explosive, flamable liquid, non-flamable gas, organic peroxide, oxidizer, and radioactive.</li> <li>5 camera angles: Azimuths of (roughly) -45°, -30°, 0°, 30°, and 45°</li> <li>5 visual field positions: top left, top right, center, bottom left, and bottom right.</li> </ul> <p>While all images are of high resolution (5184 x 3456), they are subject to different realistic lighting problems: The images with woodchip wallpaper and OSB wood backgrounds are lit with incandescent and indirect light (indoor conditions). Consequently the targets are slightly blurry (especially the peripheral ones). The images with the brick wall background are subject to harsh sunlight and sharp shadows (see top row in the figure).</p> <p>We developed the dataset for Mohamed, Tünnermann, and Mertsching (2018) to evaluate attention-accelerated hazmat label detection. However, it may be useful for many other image processing tasks that work on hazmat labels. </p> <p>If you use this dataset in your research, please cite:</p> <p>M. Mohamed, J. Tünnermann, and B. Mertsching. Seeing Signs of Danger: Attention-Accelerated Hazmat Label Detection. In: IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR), Philadelphia, PA, USA, 2018. </p> <p>BibTeX:</p> <pre class="highlight"><code>@inproceedings {mtm2018, author = { Mahmoud Mohamed and Jan T{\&quot;u}nnermann and B{\&quot;a}rbel Mertsching }, title = { Seeing Signs of Danger: Attention-Accelerated Hazmat Label Detection }, month = { June }, year = { 2018 }, address = { Philadelphia, PA, USA }, booktitle = { 2018 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR) }, publisher = { IEEE } }</code></pre>
OSF does not support the use of Internet Explorer. For optimal performance, please switch to another browser.
Accept
This website relies on cookies to help provide a better user experience. By clicking Accept or continuing to use the site, you agree. For more information, see our Privacy Policy and information on cookie use.
Accept
×

Start managing your projects on the OSF today.

Free and easy to use, the Open Science Framework supports the entire research lifecycle: planning, execution, reporting, archiving, and discovery.