adopd/YOLOv12-Seg-X-segmentation-ADOPD

YOLOv12-Seg-X Segmentation ADOPD

Thinking with Anchors Project | ADOPD 2026 Paper: Thinking with Anchors: Grounded and Efficient Document Reasoning | ADOPD 2024 Paper | Dataset | Code

Use Restrictions

Non-commercial research use only. The ADOPD fine-tuned checkpoint weights in this repository are provided solely for non-commercial research. Commercial use of these checkpoint weights is not permitted. Users must also comply with every applicable upstream license and acceptable-use term; see USE_RESTRICTIONS.md.

Model Overview

• Model developer: Thinking with Anchors project contributors
• Base architecture: YOLOv12
• Architecture variant: YOLOv12-Seg-X
• Task: class-agnostic document entity segmentation
• Fine-tuning dataset: adopd/adopd2026
• Input: one RGB document image
• Output: entity boxes, confidence scores, and instance masks

Description

YOLOv12-Seg-X Segmentation ADOPD is a YOLOv12 instance-segmentation model fine-tuned to identify visual entities in document pages. It predicts a single entity class.

Training Data

The checkpoint was fine-tuned on polygon annotations stored in:

human_annotated_masks[].polygons

YOLO segmentation labels cannot represent disconnected components as one instance. The released exporter therefore emits one training instance per polygon component while reporting every expanded source mask.

Checkpoint Format

model.pt is a native Ultralytics checkpoint produced with the pinned YOLOv12 source. That source includes the YOLOv12 A2C2f module; unrelated or older Ultralytics installations may not deserialize the architecture.

Quick Start

git clone https://github.com/SichenZhu/ADOPD2026.git
cd ADOPD2026/release_code

git clone https://github.com/sunsmarterjie/yolov12.git upstream/yolov12
git -C upstream/yolov12 checkout 01a22c0603e0eaa6d9bd62120a391e744d92cea2

python -m pip install -e model_zoo/common
python -m pip install -e upstream/yolov12
python -m pip install -e model_zoo/yolov12

hf download adopd/YOLOv12-Seg-X-segmentation-ADOPD \
  --local-dir checkpoints/yolov12-seg-x

adopd-yolo-infer \
  --checkpoint checkpoints/yolov12-seg-x/model.pt \
  --image document.jpg \
  --confidence 0.25 \
  --output prediction.json

prediction.json contains pixel-space boxes, confidence scores, and masks.

Fine-Tuning And Evaluation

Prepare ADOPD2026 with adopd-yolo-prepare --task segment, then follow the training and evaluation commands in yolov12.

Limitations

The model predicts a single document-entity class rather than semantic entity types. Disconnected components from one source mask are independent YOLO instances. Thresholds may require calibration on new document domains.

License

The ADOPD fine-tuned checkpoint weights are subject to the non-commercial, research-only restriction above. The included AGPL-3.0 license documents the terms applicable to upstream YOLOv12 software; it does not replace the checkpoint-weight restriction. Use is permitted only when all applicable terms are satisfied.

Citation

Please cite the ADOPD 2026 and ADOPD 2024 papers.

@misc{zhu2026thinkingwithanchors,
  title={Thinking with Anchors: Grounded and Efficient Document Reasoning},
  author={Sichen Zhu and Yuchen Zhu and Wenzhuo Xu and Jason Kuen and Wanrong Zhu and Jing Shi and Xuan Shen and Quanyi Wang and Yiwei Wang and Yujun Cai and Bing Shuai and Qin Zhang and Yongxin Chen and Shilong Liu and Molei Tao and Jiuxiang Gu},
  year={2026}
}

@inproceedings{gu2024adopd,
  title={{ADOPD}: A Large-Scale Document Page Decomposition Dataset},
  author={Jiuxiang Gu and Xiangxi Shi and Jason Kuen and Lu Qi and Ruiyi Zhang and Anqi Liu and Ani Nenkova and Tong Sun},
  booktitle={The Twelfth International Conference on Learning Representations},
  year={2024},
  url={https://openreview.net/forum?id=x1ptaXpOYa}
}