FlowCLAS: Enhancing Normalizing Flow Via Contrastive Learning For Anomaly Segmentation
It addresses the problem of anomaly segmentation for safety-critical applications like space robotics and autonomous driving, offering a method that does not rely on inlier labels, though it builds incrementally on existing techniques.
The paper tackles anomaly segmentation in computer vision by introducing FlowCLAS, a self-supervised framework that uses vision foundation models and normalizing flow with contrastive learning, achieving significant outperformance on the ALLO benchmark for space robotics and competitive results on road anomaly benchmarks like Fishyscapes Lost&Found and Road Anomaly.
Anomaly segmentation is a valuable computer vision task for safety-critical applications that need to be aware of unexpected events. Current state-of-the-art (SOTA) scene-level anomaly segmentation approaches rely on diverse inlier class labels during training, limiting their ability to leverage vast unlabeled datasets and pre-trained vision encoders. These methods may underperform in domains with reduced color diversity and limited object classes. Conversely, existing unsupervised methods struggle with anomaly segmentation with the diverse scenes of less restricted domains. To address these challenges, we introduce FlowCLAS, a novel self-supervised framework that utilizes vision foundation models to extract rich features and employs a normalizing flow network to learn their density distribution. We enhance the model's discriminative power by incorporating Outlier Exposure and contrastive learning in the latent space. FlowCLAS significantly outperforms all existing methods on the ALLO anomaly segmentation benchmark for space robotics and demonstrates competitive results on multiple road anomaly segmentation benchmarks for autonomous driving, including Fishyscapes Lost&Found and Road Anomaly. These results highlight FlowCLAS's effectiveness in addressing the unique challenges of space anomaly segmentation while retaining SOTA performance in the autonomous driving domain without reliance on inlier segmentation labels.