Enhancing Remote Sensing Vision-Language Models for Zero-Shot Scene Classification

Karim El Khoury, Maxime Zanella, Benoît Gérin, Tiffanie Godelaine, Benoît Macq +3 more
9/1/2024
cs.CV

Official code: elkhouryk/RS-TransCLIP (67 stars, MIT, last commit 2025-10-13).

Abstract

Vision-Language Models for remote sensing have shown promising uses thanks to their extensive pretraining. However, their conventional usage in zero-shot scene classification methods still involves dividing large images into patches and making independent predictions, i.e., inductive inference, thereby limiting their effectiveness by ignoring valuable contextual information. Our approach tackles this issue by utilizing initial predictions based on text prompting and patch affinity relationships from the image encoder to enhance zero-shot capabilities through transductive inference, all without the need for supervision and at a minor computational cost. Experiments on 10 remote sensing datasets with state-of-the-art Vision-Language Models demonstrate significant accuracy improvements over inductive zero-shot classification. Our source code is publicly available on Github: https://github.com/elkhouryk/RS-TransCLIP

View on arXivView PDF

Code Implementations(1)

elkhouryk/RS-TransCLIPOfficial100%
MIT

[ICASSP 2025] Open-source code for the paper "Enhancing Remote Sensing Vision-Language Models for Zero-Shot Scene Classification"

673PythonAug 15, 20241 years agoMIT
aerial-imageryearth-observationimage-classificationremote-sensingsatellite-imagery+4 more

Similar Papers

Few-Shot Adaptation Benchmark for Remote Sensing Vision-Language Models

88% match1 repo192025

Few-Shot Remote Sensing Image Scene Classification with CLIP and Prompt Learning

87% match2025

SenCLIP: Enhancing zero-shot land-use mapping for Sentinel-2 with ground-level prompting

86% match1 repo122024

ZoRI: Towards Discriminative Zero-Shot Remote Sensing Instance Segmentation

86% match1 repo412024

RS-OOD: A Vision-Language Augmented Framework for Out-of-Distribution Detection in Remote Sensing

84% match2025

Harnessing Massive Satellite Imagery with Efficient Masked Image Modeling

84% match1 repo1352024

Cite this paper

@article{khoury2024enhancing,
  title  = {Enhancing Remote Sensing Vision-Language Models for Zero-Shot Scene Classification},
  author = {Karim El Khoury and Maxime Zanella and Benoît Gérin and Tiffanie Godelaine and Benoît Macq and Saïd Mahmoudi and Christophe De Vleeschouwer and Ismail Ben Ayed},
  year   = {2024},
  eprint = {2409.00698},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2409.00698v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2409.00698.svg)](https://codeofpaper.com/paper/2409.00698)

Discussion