Big Data Analytics Tool

drone, big data, segmentation, AdaBoost, WaldBoost,Random Forests,CNN

Big Data analytics tool is system for object detection, classification, and segmentation utilising custom raw data as well as annotated external datasets. By utilising a deep segmentation network trained on the Semantic Drone Dataset, the tool enables the processing of Big Data in the form of raw video from drones and associated metadata. For any given scene, the information is aggregated, fed into the training pipeline, and leads to the creation of a segmentation model. By providing per-pixel class information, the outputs can be aggregated to an image-wide or scene-wide categorization, enabling further information extrapolation from the data. The tool consists of object detection, extraction of objects from video/images, storage of the images and processing of a data base of the objects, (semi) automatic annotation of the object data base, and analytics/machine learning subtools for processing of the data base (AdaBoost/WaldBoost and/or Random Forests and/or CNN based tools, etc.). 

Fakulta informačních technologií VUT v Brně, Božetěchova 2, 612 66 Brno, Q301