GeoComp demos

Object recognition with deep learning

We proposed in [1] an integrated object detection and classification pipeline using deep learning techniques for extracting various object categories in continuously streamed LIDAR point clouds collected from urban areas.

A public benchmark database called SZTAKI-Velo64Road is also released for evaluating object recognition algorithms in urban environments based on real time Lidar measurements of a Velodyne HDL 64-E sensor.

3D WebGL demo on object detection and classifaction

Object detection and recognition in a street scenario with four object classes: pedestrian, street clutter, vehicle and short facade .