Contamination MapVision / Model
AnomalyGPT, supervised (demo model)
- Organisation
- Institute of Automation, Chinese Academy of Sciences / Objecteye / Wuhan AI Research
- Family
- AnomalyGPT · version train_supervised
- Released
- 2023-08-29 (arXiv v1 submission date (https://arxiv.org/abs/2308.15366); release date of this checkpoint not stated)
- Checkpoints
- FantasticGNU/AnomalyGPT train_supervised/pytorch_model.pt
- Weights
- Open
- Links
- PaperCode
Training data disclosure: Partial list
Level determined for 37% of benchmark cells.
- TestTest data seen0
- SeenSeen, split unclear3
- TrainTraining data seen0
- Related data seen0
- NoNot seen0
- ?Unknown5
“Supervised on MVTec-AD, VisA, MVTec-LOCO-AD and CrackForest”
“we use the supervised setting as our default model to attain an enhanced user experience.”
Other declared training data
-
MVTec AD · split: unknown · fine-tuning
Split not stated. 'Supervised' training uses anomalous images with masks; in MVTec AD, anomalous images and their ground_truth masks exist only in the test split, and the dataset code derives mask paths by replacing 'test' with 'ground_truth', so test images are very likely included.
Evidence
“Supervised on MVTec-AD, VisA, MVTec-LOCO-AD and CrackForest”
raw.githubusercontent.com · readme · 2.4 weights table · retrieved 2026-10-15 “mask_path = img_path.replace('test', 'ground_truth')”
raw.githubusercontent.com · code · code/datasets/all_supervised_with_cn.py · retrieved 2026-10-15 -
VisA · split: unknown · fine-tuning
Evidence
“Supervised on MVTec-AD, VisA, MVTec-LOCO-AD and CrackForest”
raw.githubusercontent.com · readme · 2.4 weights table · retrieved 2026-10-15 -
MVTec LOCO AD · split: unknown · fine-tuning
Anomalous MVTec LOCO images are only in its test split; split not stated.
Evidence
“Supervised on MVTec-AD, VisA, MVTec-LOCO-AD and CrackForest”
raw.githubusercontent.com · readme · 2.4 weights table · retrieved 2026-10-15 -
CrackForest · split: unknown · fine-tuning
Evidence
“Supervised on MVTec-AD, VisA, MVTec-LOCO-AD and CrackForest”
raw.githubusercontent.com · readme · 2.4 weights table · retrieved 2026-10-15 -
pre-training data of PandaGPT (pandagpt4_visual_instruction_data) (not yet matched to a dataset) · split: unknown · fine-tuning
PandaGPT visual instruction data (HF dataset openllmplayground/pandagpt_visual_instruction_dataset); the paper says LLaVA/PandaGPT were pre-trained on 160k images with multi-turn conversations.
Evidence
“We perform alternating training using both the pre-training data of PandaGPT [25] and our anomaly image-text data.”
arxiv.org · paper · Section 4, Implementation Details · retrieved 2026-10-15 -
Pre-training data of base model imagebind-huge (not tracked here) (broad web data: may contain benchmark test sets) · split: unknown
Evidence
“We utilize ImageBind-Huge [8] as the image encoder and Vicuna-7B [3] as the inferential LLM, connected through a linear layer.”
arxiv.org · paper · Section 4, Implementation Details · retrieved 2026-10-15 -
Pre-training data of base model vicuna-7b (not tracked here) (broad web data: may contain benchmark test sets) · split: unknown
Evidence
“We utilize ImageBind-Huge [8] as the image encoder and Vicuna-7B [3] as the inferential LLM, connected through a linear layer.”
arxiv.org · paper · Section 4, Implementation Details · retrieved 2026-10-15 -
Pre-training data of base model pandagpt-7b (not tracked here) (broad web data: may contain benchmark test sets) · split: unknown
Evidence
“We initialize our model using pre-trained parameters from PandaGPT [25].”
arxiv.org · paper · Section 4, Implementation Details · retrieved 2026-10-15 “In our experiments and online demo, we use the Vicuna-7B and `openllmplayground/pandagpt_7b_max_len_1024` due to the limitation of computation resource.”
raw.githubusercontent.com · readme · 2.4 · retrieved 2026-10-15
Benchmark cells
Industrial anomaly detection
Seen, split unclear MVTec AD · in fine-tuning data: the dataset was used; the source does not say which part
Evidence (2)
“Supervised on MVTec-AD, VisA, MVTec-LOCO-AD and CrackForest”
“mask_path = img_path.replace('test', 'ground_truth')”
Unknown MVTec AD 2 · training data not disclosed in enough detail
Seen, split unclear MVTec LOCO AD · in fine-tuning data: the dataset was used; the source does not say which part
Evidence (1)
“Supervised on MVTec-AD, VisA, MVTec-LOCO-AD and CrackForest”
Seen, split unclear VisA · in fine-tuning data: the dataset was used; the source does not say which part
Evidence (1)
“Supervised on MVTec-AD, VisA, MVTec-LOCO-AD and CrackForest”
Unknown Real-IAD · training data not disclosed in enough detail
Unknown MPDD · training data not disclosed in enough detail
Unknown BTAD · training data not disclosed in enough detail
Unknown KolektorSDD2 · training data not disclosed in enough detail
Notes
Not used for paper results; default model of the online demo. Partial list: the supervised dataset code also has branches for 'crack_road', 'iva_road' and 'Magnetic-Tile-Defect' images not named in the README, and the splits used are not stated.
Sources
- AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models (AAAI 2024) · paper · retrieved 2026-10-15
- AnomalyGPT README (repo now CASIA-LMC-Lab/AnomalyGPT) · readme · retrieved 2026-10-15
- code/datasets/all_supervised_with_cn.py · code · retrieved 2026-10-15