Enhancing Remote Sensing Vision-Language Models for Zero-Shot Scene Classification
This work addresses a specific bottleneck in remote sensing applications by enhancing zero-shot classification without supervision, though it is incremental as it builds on existing vision-language models.
The paper tackles the problem of limited effectiveness in zero-shot scene classification for remote sensing vision-language models by introducing a transductive inference method that leverages initial predictions and patch affinity relationships, achieving significant accuracy improvements across 10 datasets.
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