Malicious Prompt Detection on GPT4V-Caption (IOD)
0FPRVLMShield
Evaluation Results
| Method | Links | |
|---|---|---|
| VLMShieldTarget Model=LLaVA-1.5-13B2026.04 | 0 | |
| VLMShieldTarget Model=Qwen2.5-VL-7B-Instruct2026.04 | 0 | |
| CIDERTarget Model=LLaVA-1.5-13B2026.04 | 2.2 | |
| CIDERTarget Model=Qwen2.5-VL-7B-Instruct2026.04 | 2.2 | |
| ASTRATarget Model=Qwen2.5-VL-7B-Instruct2026.04 | 2.26 | |
| JailGuardTarget Model=Qwen2.5-VL-7B-Instruct2026.04 | 2.64 | |
| VLMGuardTarget Model=Qwen2.5-VL-7B-Instruct2026.04 | 2.67 | |
| ECSOTarget Model=Qwen2.5-VL-7B-Instruct2026.04 | 3.7 | |
| ASTRATarget Model=LLaVA-1.5-13B2026.04 | 3.85 | |
| VLMGuardTarget Model=LLaVA-1.5-13B2026.04 | 4.76 | |
| JailGuardTarget Model=LLaVA-1.5-13B2026.04 | 4.91 | |
| ECSOTarget Model=LLaVA-1.5-13B2026.04 | 6.02 | |
| MirrorCheckTarget Model=LLaVA-1.5-13B2026.04 | 7.94 | |
| MirrorCheckTarget Model=Qwen2.5-VL-7B-Instruct2026.04 | 7.94 |