IV 2022 BSL Workshop: Beyond Supervised Learning: Addressing Data Scarcity in Intelligent Transportation Systems

June 5th, 2022 in Aachen, Germany


The workshop “Beyond Supervised Learning: addressing data scarcity in Intelligent Transportation Systems” wishes to bring together a diverse group of autonomous driving and computer vision researchers and practitioners from industry and academia to discuss novel ways to solve the data-scarcity problems in intelligent vehicles. Specifically, the workshop targets emerging new methods easing the burden of expensive manual labelling through (i) methods for learning with limited labeled data (few-shot learning, weak supervision), (ii) adaptation to novel concepts and data appearances without extensive manual re-labelling (domain adaptation, incremental and open-world recognition), and (iii) ways of economic data acquisition (data augmentation, few-click and interactive annotations).


Authors are invited to submit papers within the scope of intelligent vehicles. Topics of interest include but are not limited to the following:

  • Perception outside the vehicle with limited training data: semantic segmentation, classification, object detection, instance segmentation, panoptic segmentation, depth estimation, flow estimation, and maneuver prediction with limited training data

  • Recognition inside the vehicle with limited training data: driver state monitoring and behavior understanding with limited training data

  • Data-efficient learning: semi-, few-shot and zero-shot recognition in the context of ITS

  • Learning with partial annotations: weakly-, self-, omni-, and unsupervised learning for ITS

  • Methods for mitigating imperfect labels and data noise

  • Transfer learning, domain adaptation and generalization, and knowledge distillation for ITS

  • Open world recognition: identifying and handling novel concepts and anomalous events

  • Incremental learning, active learning, and continual learning for ITS

  • Economic data acquisition: interactive few-click labelling, data augmentation

  • Generating and learning from synthetic data and simulations