Xu Si¹  Xinming Wu^1,†,‡  Hanlin Sheng¹ 
Jun Zhu¹  Zefeng Li¹ 

¹ University of Science and Technology of China 
^† Corresponding Author  ^‡ Project Lead 

---

🌟 Spec-based Foundation Model Supports A Wide Range of Seismology

As shown in this figure, SeisCLIP can provide services for downstream tasks including event classification 💥 , location 🌍 , mechanism ⛰, etc.

🌟 News

• 2024.2.2: 🌟🌟🌟 Congratulation! The paper has been published on IEEE Transactions on Geoscience and Remote Sensing (IEEE TGRS) Links.
• 2023.9.14: 🌟🌟🌟 Pretrained weight and a simple usage demo for out SeisCLIP have been released. The implementation of SeisCLIP for event classification also released. Because the location and focal mechanism analysis code need lib 'Pytorch_geometric', it may be challenging for beginners. To provide a more detailed documentation, we will release it later. (Python Version 3.9.0 is recommended)
• 2023.9.8: Paper is released at arxiv, and code will be gradually released.
• 2023.8.7: Github Repository Initialization. (copy README template from Meta-Transformer)

🔓 Model Zoo

Open-source Modality-Agnostic Models

Model	Pretraining	Spec Size	#Param	Download	国内下载源
SeisCLIP	STEAD-1M	50 × 120	-	ckpt	[ckpt]
SeisCLIP	STEAD-1M	50 × 600	-	ckpt	[ckpt]

 

Citation

If the code and paper help your research, please kindly cite:

@ARTICLE{
  author={Si, Xu and Wu, Xinming and Sheng, Hanlin and Zhu, Jun and Li, Zefeng},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={SeisCLIP: A Seismology Foundation Model Pre-Trained by Multimodal Data for Multipurpose Seismic Feature Extraction}, 
  year={2024},
  volume={62},
  pages={1-13},
  doi={10.1109/TGRS.2024.3354456}}

License

This project is released under the MIT license.

Acknowledgement

This code is developed based on excellent open-sourced projects including CLIP, OpenCLIP, AST, MetaTransformer, ViT-Adapter, Seisbench, STEAD and PNW.