Chemist Eye: Experimental Assets for Safety Monitoring in Self-Driving Laboratories

This Zenodo repository contains CAD files, ROS bag recordings, and image datasets used in the Chemist Eye system for laboratory safety monitoring and robot decision-making. The assets support full reproducibility of the experiments presented in the associated publication. ExtendedDataset contains 900 images categorised in three clases, NORMAL, PPE and PRONE as well as the python scripts necessary to reproduce the results, each experiment contains the images classified automatically by the VLM and YOLOpose. The images extend the original Dataset by including multiple people and oclussions.本Zenodo仓库存储了用于Chemist Eye系统的计算机辅助设计（CAD）文件、机器人操作系统（ROS）包录制数据与图像数据集，该系统面向实验室安全监测与机器人决策任务开发。本仓库内的全部资源可完整复现关联学术论文中呈现的实验结果。扩展数据集（ExtendedDataset）包含900张图像，分为NORMAL、PPE（个人防护装备，Personal Protective Equipment）、PRONE三类，同时附带复现实验结果所需的Python脚本。各实验均配有经视觉语言模型（Vision-Language Model，VLM）与YOLOpose自动分类的图像。该图像数据集通过纳入多人物场景与遮挡场景，对原始数据集进行了扩充。