It is important to analyze the context to understand the activities and actions that are happening. As part of this project, several new efficient algorithms for semantic scene segmentation from different modalities have been developed, within the novel proposed MiniNet architecture. The main results have been published here:
3D-MiniNet: Learning a 2D Representation from Point Clouds for Fast and Efficient 3D LIDAR Semantic Segmentation. I. Alonso, Luis Riazuelo, L. Montesano, Ana C. Murillo. IEEE Robotics and Automation Letters (RA-L), 2020.
MiniNet: An Efficient Semantic Segmentation ConvNet for Real-time Robotic Applications. I. Alonso, L. Riazuelo, Ana C. Murillo. IEEE Transactions on Robotics (T-RO), 2020.
More details can be found here: https://sites.google.com/a/unizar.es/semanticseg/