Skip to content
/ MSA Public

Code of paper “Transcending Fusion A Multi Scale Alignment Method for Remote Sensing Image Text Retrieval”

License

Notifications You must be signed in to change notification settings

yr666666/MSA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MSA

Code of paper “Transcending Fusion: A Multi Scale Alignment Method for Remote Sensing Image Text Retrieval” Arxiv:https://arxiv.org/pdf/2405.18959

Preparation:

  1. Setup Environment
  2. Update the data path in option/RSITMD_AMFMN.yaml to match your own data location.
  3. Modify the path for the ResNet weights on line 250 of resnet.py. Download link for ResNet weights: ResNet50: https://download.pytorch.org/models/resnet50-19c8e357.pth
  4. You will need to modify the path for /bert-base-uncased/ on line 482 of retrieval.py and line 225 of model.py. /bert-base-uncased/ can be downloaded from the Hugging Face official website.
  5. Set batch_size in option/RSITMD_AMFMN.yaml. Training epoch and learning rate can be modified on lines 480 and 481 of retrieval.py.

Execution: Run the following command: python retrieval.py

This is running convergence curve: image

About

Code of paper “Transcending Fusion A Multi Scale Alignment Method for Remote Sensing Image Text Retrieval”

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages