Sit Back and Relax: Learning to Drive Incrementally in All Weather Conditions
This addresses the challenge of maintaining reliable autonomous driving systems across all weather conditions, though it is incremental as it builds on existing domain-incremental learning approaches.
The paper tackles the problem of object detection models in autonomous driving deteriorating in varying weather conditions and suffering from catastrophic forgetting when adapted sequentially, proposing a domain-incremental learning method that achieves robust performance with lightweight memory storage of less than 2% of model parameters and automatic weather inference without labels at test time.
In autonomous driving scenarios, current object detection models show strong performance when tested in clear weather. However, their performance deteriorates significantly when tested in degrading weather conditions. In addition, even when adapted to perform robustly in a sequence of different weather conditions, they are often unable to perform well in all of them and suffer from catastrophic forgetting. To efficiently mitigate forgetting, we propose Domain-Incremental Learning through Activation Matching (DILAM), which employs unsupervised feature alignment to adapt only the affine parameters of a clear weather pre-trained network to different weather conditions. We propose to store these affine parameters as a memory bank for each weather condition and plug-in their weather-specific parameters during driving (i.e. test time) when the respective weather conditions are encountered. Our memory bank is extremely lightweight, since affine parameters account for less than 2% of a typical object detector. Furthermore, contrary to previous domain-incremental learning approaches, we do not require the weather label when testing and propose to automatically infer the weather condition by a majority voting linear classifier.