Abstract: Vision-language foundation models (VLMs) have shown great potential in feature transfer and generalization across a wide spectrum of medical-related downstream tasks. However, fine-tuning ...
Abstract: Most existing adversarial attack methods for remote sensing images merely add adversarial perturbations or patches, resulting in visually unnatural modifications. Clouds are common ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results