Home

Awesome

<div align="center"> <img src="./figures/Logo.svg" width="100%"> </div>

CrackMamba

This repository is the official PyTorch implementation of Mamba meets crack segmentation.

🍓 Paper:

📢 Introduction

This repository mainly includes 2 parts:

📌 Updates

🏎️ Getting Started

1. Requirements

Our version requirements are lower than those of other Mamba-based research, which is more friendly to those who installed old versions.

Recommended versions are
    * Mamba = 1.1.1 or 1.2.0
    * causal-conv1d = 1.1.1
    * python = 3.8.15
    * pytorch = 1.12.1
    * CUDA 11.6.2 and CUDNN 8.6.0  

I want to let you know that the following installation works in Linux (specifically, I used Ubuntu 18.04). It does not work when I try to reproduce it in Win 11. 😤

2. Installation of Mamba

Step 1: Install causal_conv 1.1.1

pip install causal_conv1d==1.1.1

The screenshot of the successful installation is shown below.

<div align="center"> <img src="./figures/causal_conv1d 1.1.1.png" width="100%"> </div>

Step 2: Install Mamba 1.2.0

cd mamba-1.2.0
python setup.py install

Remark: We have provided the needed Mamba-1.2.0 in this repo for your convenience.
The screenshot of the successful installation is shown below.

<div align="center"> <img src="./figures/Mamba 1.2.0.png" width="100%"> </div>

Step 3: Install Mamba 1.2.0

💗 Cite CrackMamba

💘 Acknowledgements