Multi-source domain adaptation (MSDA) plays an important role in industrial model generalization. Recent efforts regarding MSDA focus on enhancing multi-domain distributional alignment while omitting three issues, e.g., the class-level discrepancy quantification, the unavailability of noisy pseudo labels, and source transferability discrimination, potentially resulting in suboptimal adaption performance. Therefore, we address these issues by proposing a prototype aggregation method that models the discrepancy between source and target domains at the class and domain levels. Our method achieves domain adaptation based on a group of prototypes (i.e., representative feature embeddings). A similarity score-based strategy is designed to quantify the transferability of each domain. At the class level, our method quantifies class-specific cross-domain discrepancy according to reliable target pseudo labels. At the domain level, our method establishes distributional alignment between noisy pseudo-labeled target samples and the source domain prototypes. Therefore, adaptation at the class and domain levels establishes a complementary mechanism to obtain accurate predictions. The results on three standard benchmarks demonstrate that our method outperforms most state-of-the-art methods. In addition, we provide further elaboration of the proposed method in light of the interpretable results obtained from the analysis experiments.

Overhead transmission lines are important lifelines in power systems, and the research and application of their intelligent patrol technology is one of the key technologies for building smart grids. The main reason for the low detection performance of fittings is the wide range of some fittings’ scale and large geometric changes. In this paper, we propose a fittings detection method based on multi-scale geometric transformation and attention-masking mechanism. Firstly, we design a multi-view geometric transformation enhancement strategy, which models geometric transformation as a combination of multiple homomorphic images to obtain image features from multiple views. Then, we introduce an efficient multiscale feature fusion method to improve the detection performance of the model for targets with different scales. Finally, we introduce an attention-masking mechanism to reduce the computational burden of model-learning multiscale features, thereby further improving model performance. In this paper, experiments have been conducted on different datasets, and the experimental results show that the proposed method greatly improves the detection accuracy of transmission line fittings.

Multi-source domain adaptation (MSDA) plays an important role in industrial model generalization. Recent efforts on MSDA focus on enhancing multi-domain distributional alignment while omitting three issues, e.g., the class-level discrepancy quantification, the unavailability of noisy pseudo-label, and source transferability discrimination, potentially resulting in suboptimal adaption performance. Therefore, we address these issues by proposing a prototype aggregation method that models the discrepancy between source and target domains at the class and domain levels. Our method achieves domain adaptation based on a group of prototypes (i.e., representative feature embeddings). A similarity score-based strategy is designed to quantify the transferability of each domain. At the class level, our method quantifies class-specific cross-domain discrepancy according to reliable target pseudo-labels. At the domain level, our method establishes distributional alignment between noisy pseudo-labeled target samples and the source domain prototypes. Therefore, adaptation at the class and domain levels establishes a complementary mechanism to obtain accurate predictions. The results on three standard benchmarks demonstrate that our method outperforms most state-of-the-art methods. In addition, we provide further elaboration of the proposed method in light of the interpretable results obtained from the analysis experiments.