Images Classification

The collection of images has classification labels for each images for the eight classes.

Object Detection

Each object has bounding boxes marking the location of the object in the image.

Semantic Segmentation

A label image contains the pixel-wise semantic segmentation of the objects conatined in an image.

General information

FireNet is an open ML training dataset for visual recognition of fire safety equipment. Our dataset directly links the objects to their respective Uniclass, the construction sector’s classification scheme to name objects. FireNet has been designed as a ML training dataset for experimentation and therefore fulfills multiple machine learning scenarios (classification, object detection, semantic segmentation). It is intended as a domain specific dataset to refine pre-trained standard architectures.

Future iterations

The labelling of the images (delineating the objects in the images) was crowd sourced. We used the dominating Amazon Mechanical Turk (MTurk) platform for this task. All crowd sourced labels were verified in-house and accepted or rejected accordingly. As part of a continuous maintenance to the dataset we are exploring options to revisit rejected images. If you are interested in sponsoring further classes or images for FireNet please contact us using the details at the bottom of this page.