Towards generalization on real domain for single image dehazing via meta-learning

Abstract

Learning-based image dehazing methods are essential to assist autonomous systems in enhancing reliability. Due to the domain gap between synthetic and real domains, the internal information learned from synthesized images is usually sub-optimal for real domains. Thus, the dehazing models trained on synthetic images often suffer from severe performance drops on real-world samples. Driven by the ability to explore internal information from a few unseen-domain samples, meta-learning is commonly adopted to address this issue via test-time training, which is hyperparameter-sensitive and time-consuming. In contrast, we present a domain generalization framework based on meta-learning to dig out representative and discriminative internal properties of real hazy domains without test-time training. To obtain representative domain-specific information, we attach two entities termed adaptation network and distance-aware aggregator to our dehazing network. The adaptation network assists in distilling domain-relevant information from a few hazy samples and caching it into a collection of features. The distance-aware aggregator strives to summarize the generated features and filter out misleading information for more representative internal properties. To enhance the discrimination of distilled internal information, we present a novel loss function called domain-relevant contrastive regularization. It encourages the similarity of the internal features generated from the same domain and the distinction of the features from diverse domains. The generated representative and discriminative features are regarded as some external variables of our dehazing network to regress a particular function for a given domain. The extensive experiments on real hazy datasets validate that our proposed method has superior generalization ability than the state-of-the-art competitors.