The Computer Vision Center (CVC) is a non-profit research center established in 1995 by the Generalitat de Catalunya and the Universitat Autònoma de Barcelona (UAB). Its mission is to carry out cutting-edge research that has the highest international impact in the field of computer vision. It also promotes the transference of knowledge to industry and society.
One of the fields in which CVC works is Autonomous Driving (AD) enabled by Deep Learning (DL). A current bottleneck is the relative low availability of annotated data to train artificial drivers based on DL. Thus, a line of work at CVC is to automatically generate photo-realistic scenarios by means of Computer Graphics, and, more specifically by using professional frameworks which support this task (e.g., Houdini, Unity3D/UE4, Octane, etc.).
The project Automatic Data Annotation by Semi-Supervised Learning and Procedural Generation of Augmented Scenarios (ADA, lead by Dr. Antonio M. López, and funded by Spanish Ministerio de Ciencia e Innovación & Agencia Estatal de Investigación, studies the use of photo-realistic images and SSL to train deep vision models for autonomous driving tasks, removing the synth-to-real domain gap along the way. In this context, focusing on on-board image generation with automatically ground truth, we look for a junior researcher (R1) – especially in the last year of their PhD - to collaborate in the project, focusing his research on developing software pipelines which minimize the human annotation effort required to provide ground truth (annotations) for sensor data acquired on-board AVs, so that deep perception models can be trained and validated in a more cost-effective manner; in the best case, even avoiding human annotation.
Més posts de Recerca
Comentaris