Anomaly Detection on TAD (test)
97.5Overall AUCLLaVA-1.6 SKR
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LLaVA-1.6 SKRModel=LLaVA-1.6, Adaptation Method=SKR, Runtime (min/100 cases)=1.32026.04 | 97.5 | — | 87.9 | |
| Qwen2-VL SKRModel=Qwen2-VL, Adaptation Method=SKR, Runtime (min/100 cases)=0.72026.04 | 96 | — | 86.8 | |
| Qwen2-VL SFTModel=Qwen2-VL, Adaptation Method=SFT, Runtime (min/100 cases)=0.92026.04 | 96 | — | 74.6 | |
| Phi-3-VL SFTModel=Phi-3-VL, Adaptation Method=SFT, Runtime (min/100 cases)=7.12026.04 | 94.5 | — | 85 | |
| Phi-3-VL SKRModel=Phi-3-VL, Adaptation Method=SKR, Runtime (min/100 cases)=0.72026.04 | 93.5 | — | 93.3 | |
| Qwen2-VL BaseModel=Qwen2-VL, Adaptation Method=Base, Runtime (min/100 cases)=1.12026.04 | 93.5 | — | 53.2 | |
| LLaVA-1.6 BaseModel=LLaVA-1.6, Adaptation Method=Base, Runtime (min/100 cases)=3.82026.04 | 93 | — | 53.1 | |
| UMILCategory=WSVAD2023.03 | 92.93 | 65.82 | — | |
| WSALSupervision setting=Weakly-supervised2020.08 | 89.64 | 61.66 | — | |
| WSALCategory=WSVAD2023.03 | 89.64 | 61.66 | — | |
| BaselineCategory=WSVAD2023.03 | 89.1 | 56.47 | — | |
| GIGCategory=WSVAD2023.03 | 85.64 | 58.65 | — | |
| LLaVA-1.6 SFTModel=LLaVA-1.6, Adaptation Method=SFT, Runtime (min/100 cases)=1.42026.04 | 83.5 | — | 38.1 | |
| Zhu et al.Supervision setting=Weakly-supervised2020.08 | 83.08 | 56.89 | — | |
| Motion-AwareCategory=WSVAD2023.03 | 83.08 | 56.89 | — | |
| Sultani et al.Supervision setting=Weakly-supervised2020.08 | 81.42 | 55.97 | — | |
| Sultani et al.Category=WSVAD2023.03 | 81.42 | 55.97 | — | |
| Phi-3-VL BaseModel=Phi-3-VL, Adaptation Method=Base, Runtime (min/100 cases)=4.72026.04 | 77 | — | 28.7 | |
| Liu et al.Supervision setting=Trained with normal videos only2020.08 | 69.13 | 55.38 | — | |
| Liu et al.Category=UVAD2023.03 | 69.13 | 55.38 | — | |
| Luo et al.Supervision setting=Trained with normal videos only2020.08 | 57.89 | 55.84 | — | |
| Luo et al.Category=UVAD2023.03 | 57.89 | 55.84 | — | |
| SVM BaselineCategory=UVAD2023.03 | 50 | 50 | — |