Online 3D object tracking with low-cost depth cameras (2015)
This is the supplementary video for paper:
D. Tan, F. Tombari, S. Ilic, N. Navab, “A Versatile Learning-based 3D Temporal Tracker: Scalable, Robust, Online”, ICCV 2015
3D Head Pose Estimation with Random Forests (2015)
This is the supplementary video for paper:
D. Tan, F. Tombari, N. Navab, “A Combined Generalized and Subject-Specific 3D Head Pose Estimation”, 3DV 2015
Real-time and scalable incremental segmentation (2015)
This is the supplementary video for paper:
K. Tateno, F. Tombari, N. Navab, “Real-Time and Scalable Incremental Segmentation on Dense SLAM”
BOLD features for texture-less object detection (2013)
In this video we compare BOLD features with other state-of-the-art descriptors within both textureless and textured object detection. BOLD features were proposed in
F. Tombari, A. Franchi, L. Di Stefano, “BOLD features for texture-less object detection”, ICCV 2013
RGB-D Object Recognition and Pose Estimation based on multiple features (2013)
This video demonstrates the use of Global Hypothesis Verification as well as the use of multiple features (3D local, 3D global, 2D local) for Object Recognition in RGB-D within clutter and occlusions. The video shows the performance of the method proposed in
A. Aldoma, F. Tombari, J. Prankl, A. Richtsfeld, L. Di Stefano, M. Vincze, “Multimodal Cue Integration through Hypotheses Verification for RGB-D Object Recognition and 6DOF Pose Estimation”, ICRA 2013
on a public dataset concerning typical household objects and high levels of occlusions
Semantic Segmentation of RGB-D data (2011)
These 2 videos demonstrate semantic segmentation of RGB-D data (color + depth). In these experiments data has been acquired using a Microsoft Kinect sensor. Each frame is automatically segmented into 8 different object classes, which include both object categories (e.g. juice bottles) and specific object instances (e.g. Super Mario).
The reference papers describing the technique being used are:
F. Tombari, L. Di Stefano, “3D Data Segmentation by Local Classification and Markov Random Fields”, 3DIMPVT 2011
F. Tombari, L. Di Stefano, S. Giardino, “Online Learning for automatic segmentation of 3D data”, IROS 2011
3D Object Recognition and Pose Estimation on RGB-D data (2011)
This video demonstrates 3D object recognition and 3D pose estimation based on RGB-D data (color + depth). In this experiment data is acquired using a Microsoft Kinect sensor. The reference paper is:
F. Tombari, S. Salti, L. Di Stefano, “RGB-D object recognition and localization with clutter and occlusions”, RGB-D Workshop on 3D Perception in Robotics, 2011 [PDF]
3D Acquisition based on improved Space-time stereo (2010)
This video shows a single view acquisition of a face by means of the 3D acquisition technique described in:
F. Tombari, L. Di Stefano, S. Mattoccia, A.Mainetti, “A 3D reconstruction system based on improved Spacetime Stereo”, ICARCV 2010
3D Object Recognition and Pose Estimation on stereo data (2010)
Demo video based on a stereo camera performing object recognition and 3D pose estimation with clutter and occlusion using the Hough voting scheme proposed in:
F. Tombari, L. Di Stefano, “Object recognition in 3D scenes with occlusions and clutter by Hough voting”, PSIVT 2010
Depth Tracker on Icub (2009)
Demo video of a tracker implemented on the Icub based on stereo vision and a visual attention field module, developed during the VeniVidiVici RobotCub Summer School.
Multiview graffiti detection (2008)
Video showing real-time graffiti detection based on a multiview sensor (stereo camera). The technique is proposed in:
L. Di Stefano, F. Tombari, A. Lanza, S. Mattoccia, S. Monti, “Graffiti detection using two views”, VS 2008