Surface Vision Mamba (SiM) applies a bidirectional state space model to icosphere patches of cortical surfaces, matching or outperforming transformer- and GDL-based models with up to 4.8x faster inference and 91.7% lower memory.
Benchmarking geometric deep learning for cortical segmentation and neurodevel- opmental phenotype prediction
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Surface Vision Mamba: Leveraging Bidirectional State Space Model for Efficient Spherical Manifold Representation
Surface Vision Mamba (SiM) applies a bidirectional state space model to icosphere patches of cortical surfaces, matching or outperforming transformer- and GDL-based models with up to 4.8x faster inference and 91.7% lower memory.