Linking vision and motion for self-supervised object-centric perception
This work addresses the need for more generalizable perception in autonomous driving by reducing reliance on supervised data, though it is incremental as it builds on existing self-supervised object-centric methods.
The paper tackled the problem of learning object-centric representations for autonomous driving without supervision by adapting a self-supervised vision model using only RGB video and vehicle pose inputs, and demonstrated promising results on the Waymo Open dataset with successful tracking of vehicles and pedestrians.
Object-centric representations enable autonomous driving algorithms to reason about interactions between many independent agents and scene features. Traditionally these representations have been obtained via supervised learning, but this decouples perception from the downstream driving task and could harm generalization. In this work we adapt a self-supervised object-centric vision model to perform object decomposition using only RGB video and the pose of the vehicle as inputs. We demonstrate that our method obtains promising results on the Waymo Open perception dataset. While object mask quality lags behind supervised methods or alternatives that use more privileged information, we find that our model is capable of learning a representation that fuses multiple camera viewpoints over time and successfully tracks many vehicles and pedestrians in the dataset. Code for our model is available at https://github.com/wayveai/SOCS.