SPC-Verify Framework Detects Adversarial Attacks on Image Captioning Models
Researchers have introduced SPC-Verify, a dual-branch framework that detects adversarial examples in image captioning models by evaluating visual-semantic consistency and ...
