Domain Similarity-Perceived Label Assignment for Domain Generalized Underwater Object Detection
This addresses domain generalization for underwater object detection, an incremental improvement over existing domain adversarial learning methods.
The paper tackles domain shift in underwater object detection by introducing Domain Similarity-Perceived Label Assignment (DSP), which replaces one-hot domain labels with similarity-based assignments, achieving state-of-the-art results on the S-UODAC2020 benchmark and validating effectiveness on Cityscapes.
The inherent characteristics and light fluctuations of water bodies give rise to the huge difference between different layers and regions in underwater environments. When the test set is collected in a different marine area from the training set, the issue of domain shift emerges, significantly compromising the model's ability to generalize. The Domain Adversarial Learning (DAL) training strategy has been previously utilized to tackle such challenges. However, DAL heavily depends on manually one-hot domain labels, which implies no difference among the samples in the same domain. Such an assumption results in the instability of DAL. This paper introduces the concept of Domain Similarity-Perceived Label Assignment (DSP). The domain label for each image is regarded as its similarity to the specified domains. Through domain-specific data augmentation techniques, we achieved state-of-the-art results on the underwater cross-domain object detection benchmark S-UODAC2020. Furthermore, we validated the effectiveness of our method in the Cityscapes dataset.