AAGNet
A B-Rep segmentation neural network based on attributed adjacency graphs. Building on the face adjacency graph, it encodes inter-face geometric relationships—such as convex or concave transitions and shared edge types—as edge attributes, capturing finer-grained assembly and manufacturing feature semantics than standard adjacency-based approaches.
Timeline (last 90 days)
The study introduces two representative B-Rep segmentation models, UV-Net and AAGNet, and trains them on a newly constructed manually annotated benchmark dataset.
AAGNet (Attributed Adjacency Graph Network) encodes geometric relationships between faces as edge attributes of the graph, in order to capture finer-grained assembly and feature semantics.
Learning-based methods can achieve very strong performance when annotated data is sufficient, but their robustness degrades significantly when facing domain shift.
All Facts (3)
The study introduces two representative B-Rep segmentation models, UV-Net and AAGNet, and trains them on a newly constructed manually annotated benchmark dataset.
50%UnverifiedAAGNet (Attributed Adjacency Graph Network) encodes geometric relationships between faces as edge attributes of the graph, in order to capture finer-grained assembly and feature semantics.
50%UnverifiedLearning-based methods can achieve very strong performance when annotated data is sufficient, but their robustness degrades significantly when facing domain shift.
50%