Photometric Stereo
Retrieving geometric information provided from shading images is a challenging problem which has been widely studied in the CRL. Several resources are provided here in order to evaluate latest CVG methods including (i) LUCES a dataset containing PS images and ground truth meshes of 14 objects, (ii) PX-NET sample code and (iii) point-light adaptation of the PX-NET sample code.
ContextNet Segmentation
A deep neural network architecture which builds on factorized convolution, network compression and pyramid representation to produce competitive semantic segmentation in real-time with low memory requirement. ContextNet combines a deep network branch at low resolution that capture global context information efficiently with a shallow branch that focuses on high-resolution segmentation details.
Toshiba CAD dataset
CAD models of the 10 real objects used for evaluating a method for vote-based 3D shape recognition and registration, in particular using mean shift on 3D pose votes in the space of direct similarity transforms.
World Models
A world model is a representation of an environment in which an agent operates. It captures the dynamics of the environment, including how it evolves over time. It allows for sample efficiency, planning, and effective control in embodied agents.
Latest Publications
F Logothetis, I Budvytis, R Cipolla
S Morad, R Kortvelesy, S Liwicki, A Prorok,
C Zhang, S Liwicki, S He, W Smith, R Cipolla,
S Morad, R Kortvelesy, M Bettini, S Liwicki, A Prorok
