Contamination Map Methodology

Vision / Model

VCP-CLIP, trained on VisA

Organisation
Institute of Automation, Chinese Academy of Sciences et al.
Family
VCP-CLIP · version train_visa.pth
Released
2024-07-17 (arXiv v1 submission date (https://arxiv.org/abs/2407.12276))
Checkpoints
vcp_weight/train_visa.pth
Weights
Open
Links
PaperCode

Training data disclosure: Full list

Level determined for 12% of benchmark cells.

“Therefore, to evaluate ZSAS performance on other datasets, we employ weights trained on VisA's test sets.”
arxiv.org · paper · Appendix C · retrieved 2026-10-15
“"train_visa.pth" indicates that the auxiliary training dataset is VisA, which you can utilize to test any products outside of the VisA dataset, and vice versa”
raw.githubusercontent.com · readme · Prepare the pre-trained weights · retrieved 2026-10-15

Built on

Everything these base models may have seen is inherited by VCP-CLIP, trained on VisA.

Other declared training data

Benchmark cells

Industrial anomaly detection

Unknown MVTec AD · unknown for its base model OpenAI CLIP: training data not disclosed in enough detail

Unknown MVTec AD 2 · unknown for its base model OpenAI CLIP: training data not disclosed in enough detail

Unknown MVTec LOCO AD · unknown for its base model OpenAI CLIP: training data not disclosed in enough detail

Test data seen VisA · in fine-tuning data: its test split was used for training

Evidence (1)
“Therefore, to evaluate ZSAS performance on other datasets, we employ weights trained on VisA's test sets.”
arxiv.org · paper · Appendix C · retrieved 2026-10-15

Unknown Real-IAD · unknown for its base model OpenAI CLIP: training data not disclosed in enough detail

Unknown MPDD · unknown for its base model OpenAI CLIP: training data not disclosed in enough detail

Unknown BTAD · unknown for its base model OpenAI CLIP: training data not disclosed in enough detail

Unknown KolektorSDD2 · unknown for its base model OpenAI CLIP: training data not disclosed in enough detail

Notes

Frozen CLIP; learnable prompts and Pre-/Post-VCP modules. Used for all reported zero-shot segmentation results except VisA: MVTec-AD, BSD, GC, KSDD2, MSD, Road, RSDD, BTech (BTAD), DAGM.

Sources