MixerCSeg: An Efficient Mixer Architecture for Crack Segmentation via Decoupled Mamba Attention

2026-03-01

AI Summary

Existing crack segmentation methods struggle to simultaneously model local textures, global dependencies, and sequential context, resulting in inadequate representation of complex structures. This work proposes MixerCSeg, a novel architecture that unifies CNN-based local pathways, Transformer-based global pathways, and Mamba-based sequential pathways within a single encoder to achieve multidimensional feature disentanglement. The approach introduces a TransMixer mechanism to decouple Mamba’s attention behavior, complemented by direction-guided edge-gated convolution (DEGConv), spatial block processing, and a spatial refinement multi-level fusion (SRF) module. Remarkably, MixerCSeg achieves state-of-the-art performance across multiple crack segmentation benchmarks with only 2.05 GFLOPs and 2.54M parameters, demonstrating an exceptional balance between computational efficiency and representational power.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Mixture of Experts (MoE)
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Feature encoders play a key role in pixel-level crack segmentation by shaping the representation of fine textures and thin structures. Existing CNN-, Transformer-, and Mamba-based models each capture only part of the required spatial or structural information, leaving clear gaps in modeling complex crack patterns. To address this, we present MixerCSeg, a mixer architecture designed like a coordinated team of specialists, where CNN-like pathways focus on local textures, Transformer-style paths capture global dependencies, and Mamba-inspired flows model sequential context within a single encoder. At the core of MixerCSeg is the TransMixer, which explores Mamba's latent attention behavior while establishing dedicated pathways that naturally express both locality and global awareness. To further enhance structural fidelity, we introduce a spatial block processing strategy and a Direction-guided Edge Gated Convolution (DEGConv) that strengthens edge sensitivity under irregular crack geometries with minimal computational overhead. A Spatial Refinement Multi-Level Fusion (SRF) module is then employed to refine multi-scale details without increasing complexity. Extensive experiments on multiple crack segmentation benchmarks show that MixerCSeg achieves state-of-the-art performance with only 2.05 GFLOPs and 2.54 M parameters, demonstrating both efficiency and strong representational capability. The code is available at https://github.com/spiderforest/MixerCSeg.

Problem

• crack segmentation
• feature encoder
• structural representation
• complex crack patterns
• pixel-level segmentation

Innovation

• Mixer Architecture
• Decoupled Mamba Attention
• Crack Segmentation
• Edge-Guided Convolution
• Efficient Multi-Path Encoder