A New Remote Sensing Dataset for Heliport Detection

Labelled training image datasets are the main driving forces for many research areas. Especially object detection and recognition research require quite a few labelled data in order to come up with a successful model. In this work a new dataset for H-shaped heliport detection is presented. The presence of a heliport on an image usually implies an important facility such as governmental buildings or military facilities and detection of heliports can reveal critical information about the content of an image. The dataset directory and labelling structure is in Pascal Visual Object Classes (VOC) format. Hence, it is possible to use most of the well-known neural network architectures for creating a model. The dataset contains Google Maps images with heliport samples from all over the world. Therefore, the dataset represents a large variety of image context and different heliport samples.