Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
[CVPR 2025] SCSegamba: Lightweight Structure-Aware Vision Mamba for Crack Segmentation in Structures
| Date | Stars |
|---|---|
| 2026-07-24 | 282 |
| 2026-07-25 | 283 |
| 2026-07-28 | 284 |
| 2026-07-30 | 284 |
| 2026-08-06 | 284 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
<div align="center">
<h1>SCSegamba</h1>
</div>
<p align="center">
<img src="./figures/LOGO.png" alt="LOGO" width="185" height="200" />
</p>
<div align="center">
<h4>[CVPR2025] SCSegamba: Lightweight Structure-Aware Vision Mamba for Crack Segmentation in Structures</h4>
</div>
<div align="center">
<h6>🌟 If this work is useful to you, please give this repository a Star! 🌟</h6>
</div>
<div align="center">
<a href="https://arxiv.org/abs/2503.01113"><img src="https://img.shields.io/badge/Arxiv-2503.01113-b31b1b?logo=arXiv" alt="arXiv" style="height:20px;"></a>
<a href="https://www.apache.org/licenses/" style="margin-left:10px;"><img src="https://img.shields.io/badge/License-Apache%202.0-yellow" alt="License" style="height:20px;"></a>
</div>
## 📬News
- **2025-03-05**: We added a description of the the methodology and visualization of the experimental results in ReadMe💭!
- **2025-03-04**: The code for **SCSegamba** is publicly available in this repository📦!
- **2025-03-04**: The preprint of **SCSegamba** has been posted on [**📤️arXiv**](https://arxiv.org/abs/2503.01113)!
- **2025-03-01**: In the next few days, we will make some minor revisions and then publish the preprint on arXiv. The code will also be released shortly after the paper is published! **Stay tuned🥰**!
- **2025-02-27**: 🎉🎉🎉 We are thrilled to announce that our **SCSegamba** has been accepted to **CVPR 2025**!
## 🛠️ Method Overview
<p align="center">
<img src="./figures/Overview.png" alt="Overview" />
</p>
Pixel-level segmentation of structural cracks across various scenarios remains a considerable challenge. Current methods encounter challenges in effectively modeling crack morphology and texture, facing challenges in balancing segmentation quality with low computational resource usage. To overcome these limitations, we propose a lightweight Structure-Aware Vision Mamba Network (**SCSegamba**), capable of generating high-quality pixel-level segmentation maps by leveraging both the morphological information and texture cues of crack pixels with minimal computational cost. Specifically, we developed a **S**tructure-**A**ware **V**isual **S**tate Space module (**SAVSS**), which incorporates a lightweight **G**ated **B**ottleneck **C**onvolution (**GBC**) and a **S**tructure-**A**ware **S**canning **S**trategy (**SASS**). The key insight of GBC lies in its effectiveness in modeling the morphological information of cracks, while the SASS enhances the perception of crack topology and texture by strengthening the continuity of semantic information between crack pixels. Experiments on crack benchmark datasets demonstrate that our method outperforms other state-of-the-art (SOTA) methods, achieving the highest performance with only 2.8M parameters. On the multi-scenario dataset, our method reached 0.8390 in F1 score and 0.8479 in mIoU.
## 🕹Getting Started
#### Environment Setup
You can create your own conda environment for SCSegamba based on the following command⚙️:
```shell
conda create -n SCSegamba python=3.10 -y
conda activate SCSegamba
pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 -f https://download.pytorch.org/whl/torch_stable.html
pip install -U openmim
mim install mmcv-full
pip install mamba-ssm==1.2.0
pip install timm lmdb mmengine numpy
```
#### Run
You can modify the parameters in the **main.py** file and run it with the following command (The structure of the dataset folder is set to be consistent with the [TUT](https://github.com/Karl1109/TUT)):
``````shell
python main.py
``````
You can also use checkpoints for inference with the following command:
```shell
python test.py
```
Use the following commands to calculate metrics (You can find the SCSegamba test results on the TUT dataset in the `./results/results_test/TUT_results/` path and calculate the metrics using the following command.):
```shell
python eval_compute.py
cd eval
python evaluate.py
```
You can also follow the steps below to validate the results of our eExcerpt of 7,137 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:9f10931d38c4a80f, topic:deep-learning
matched fp:9f10931d38c4a80f, topic:computer-vision