README.md

metadata

license: apache-2.0
language:
  - en
task_categories:
  - visual-question-answering
pretty_name: DualXrayBench
size_categories:
  - 1K<n<10K
tags:
  - x-ray
  - dual-view
  - security-screening
  - spatial-reasoning

DualXrayBench

DualXrayBench is the 1,594-example, eight-task evaluation split accompanying the GSR work on dual-view X-ray reasoning. Each record pairs a top-view and side-view image with one multiple-choice question, its reference answer, task metadata, and reference target-instance boxes.

Files

• annotations/benchmark.jsonl: stable public records with task_id, task_name, and relative top/side image paths.
• annotations/dual-xray-benchmark.json: unchanged paper-v1 canonical compatibility file.
• annotations/tasks/*.jsonl: the same stable records partitioned by task.
• images/top and images/side: exactly 1,594 paired JPEG images in each view.
• tasks.json, schema.json, and evaluation_protocol.md: task definitions, record schema, and scoring guidance.

The record order is the frozen canonical paper-v1 order. Image paths are relative to this repository. The benchmark image IDs have zero overlap with the GSXray training image IDs used for this release.

Loading

import json

with open("annotations/benchmark.jsonl", encoding="utf-8") as handle:
    records = [json.loads(line) for line in handle]

first = records[0]
print(first["images"]["top"], first["images"]["side"], first["task_id"])

See evaluation_protocol.md before reporting results. Do not train or tune on benchmark answers.

License and attribution

This release is distributed under the Apache License 2.0 under the project's confirmed redistribution basis. It is derived from the LDXray dual-view X-ray dataset and includes GSR benchmark annotations and packaging. See LICENSE and NOTICE.

Citation

@inproceedings{peng2026secondview,
  title={Can a Second-View Image Be a Language? Geometric and Semantic Cross-Modal Reasoning for X-ray Prohibited Item Detection},
  author={Peng, Chuang and Tao, Renshuai and Ren, Zhongwei and Liu, Xianglong and Wei, Yunchao},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2026}
}

@inproceedings{tao2025dualview,
  title={Dual-view X-ray Detection: Can AI Detect Prohibited Items from Dual-view X-ray Images Like Humans?},
  author={Tao, Renshuai and Wang, Haoyu and Guo, Yuzhe and Chen, Hairong and Zhang, Li and Liu, Xianglong and Wei, Yunchao and Zhao, Yao},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={10338--10347},
  year={2025}
}