• This code is for multi-modal remote sensing image classification and natural scene parsing segmentation based on VMamba and CLIP.

Overview

Getting Started

Create the environment

Step 1: Download the M³amba repository:

To get started, first download the M³amba repository and navigate to the project directory:

download .zip file and unzip it
cd m3amba

Step 2: Environment Setup:

M³amba recommends setting up a conda environment and installing dependencies via pip. Use the following commands to set up your environment:

Create and activate a new conda environment

conda create -n m3amba python=3.8.19
conda activate m3amba

Requirments and we recommend torch 2.1.0, torchvision 0.16.0, cuda 12.1. Don't forget to follow VMamba compilation

Numpy  Matplotlib  Scipy  Tqdm  scikit-learn  thop  clip

Prepare the dataset

You can download the dataset from the reference in the paper and then run

python utils/crop.py

Begin to train and test

Modify the path in data_loader.py and train/test.py to yours, then run

python train.py
python test.py

Result

pretrained weights
test log