Project Cursa - 机器人操作视频标注质量检查
Project Cursa - Robot Manipulation Video Annotation QC
职位描述
我们正在寻找一位注重细节的质检审核员，以确保我们标注团队制作的机器人操作视频标注的质量和一致性。你将根据详细的风格指南和参考示例审核已标注的视频片段，发现错误或不一致之处，并向标注人员提供清晰、有建设性的反馈。此职位对维护用于机器人AI模型的训练数据的准确性和可靠性至关重要。
你将负责：
- 审核机器人操作视频标注的三个层级（原子动作、技能/子任务、任务/目标）的已完成标注内容。
- 核对标注人员的描述与所有三个摄像头视角（顶部和两个腕部安装的摄像头）是否一致，以验证所描述动作、物体识别和接触/抓取细节的准确性。
- 确保每个视频的完整覆盖——确认没有标注遗漏，包括空闲或暂停时刻。
- 检查描述是否准确反映视频中实际发生的内容，包括失败、掉落或滑脱的情况——标记任何结果描述不准确或被“美化”的情况。
- 评估标注是否符合风格指南：正确的词汇使用、空间/方向准确性、物体和动作的一致术语，以及适当的语言变化（避免重复表达）。
- 识别标注人员中反复出现的错误或混淆模式，并向团队负责人报告，用于校准。
- 对被拒绝或修改的标注提供具体、可操作的反馈，引用风格指南和参考示例。
- 参与团队和客户定期的校准会议，确保质检标准与不断演进的指南或边缘案例解决方案保持一致。
- 维护质检日志或跟踪表，记录审核结果、常见错误类型和标注人员表现趋势（根据项目要求）。
我们寻找的人选需满足以下条件：
- 出色的英文书面表达能力和强大的阅读理解能力——你需要评估精确、细致的语言选择，而不仅仅是检查拼写错误。
- 极强的细节关注度——能够发现视频内容与文字描述之间的细微不匹配（例如，错误标注的抓取类型、错误的物体部件、遗漏的失败事件）。
- 对空间/机械描述有扎实的理解（左右、上下、物体部件如把手、盖子、边缘），以验证标注人员的准确性。
查看英文原文
About the Role
We are looking for a detail-oriented QC Reviewer to ensure the quality and consistency of robot manipulation video annotations produced by our annotation team. You'll review labeled video segments against a detailed style guide and reference examples, catch errors or inconsistencies, and provide clear, constructive feedback to annotators. This role is critical to maintaining the accuracy and reliability of training data used for robotics AI models.
What You'll Do
- Review completed annotations of robot manipulation videos across three levels of labeling detail (atomic motion, skill/subtask, task/goal).
- Cross-check annotator descriptions against all three camera views (overhead and both wrist-mounted cameras) to verify accuracy of described actions, object identification, and contact/grasp details.
- Verify full coverage of each video — confirm there are no gaps in labeling, including idle or pause moments.
- Check that descriptions accurately reflect what actually happened in the video, including failures, drops, or slipped grips — flagging any instances where outcomes were inaccurately described or "smoothed over."
- Evaluate annotations for adherence to the style guide: correct vocabulary usage, spatial/directional accuracy, consistent terminology for objects and actions, and appropriate language variation (avoiding repetitive phrasing).
- Identify recurring errors or patterns of confusion among annotators and report them to the team lead for calibration purposes.
- Provide specific, actionable feedback on rejected or revised annotations, referencing the style guide and reference examples.
- Participate in periodic calibration sessions with the team and client to align QC standards with evolving guidelines or edge-case resolutions.
- Maintain a QC log or tracking sheet documenting review outcomes, common error types, and annotator performance trends (as required by the project).
What We're Looking For
Required:
- Excellent written English and strong reading comprehension — you'll be evaluating precise, nuanced language choices, not just checking for typos.
- Exceptional attention to detail — able to catch subtle mismatches between video content and written description (e.g., a mislabeled grasp type, an incorrect object part, a missed failure event).
- Strong grasp of spatial/mechanical description (left/right, above/below, object parts like handles, lids, edges) to verify annotator accuracy.
- Ability to apply a detailed, structured style guide consistently and make sound judgment calls in ambiguous or edge-case scenarios.
- Comfort giving clear, specific, and constructive feedback to annotators (written feedback, not just pass/fail marks).
- Reliable, self-directed work habits, with the ability to work independently while meeting review turnaround expectations.
Nice to Have:
- Prior experience in a QC, editing, proofreading, or annotation review role.
- Experience with video annotation, data labeling, or transcription QA specifically.
- Familiarity with robotics terminology (grippers, end-effectors, manipulation).
- Experience with annotation tools such as Label Studio, including review/approval workflows.
Why Join Welo Data?
✨ Limitless Flexibility
Project-based opportunities that fit your availability. Choose when and how much you want to contribute—fully remote, with complete autonomy.
🌱 Limitless Growth
Optional access to AI and Large Language Model workshops designed specifically for professionals like you. No coding required—just your expertise.
🌍 Limitless Support
Be part of a global contributor community with responsive guidance and support.
💡 Real Impact
Apply your expertise in the Legal field to influence the AI systems shaping the future of your industry—while collaborating with data professionals and expanding your skills.
How to Apply?
Apply now by answering a few quick questions to join our database and become part of our growing community.
About Welo Data
Welo Data, part of Welocalize, is a global AI data company with 500,000+ contributors delivering high-quality, ethical data to train the world’s most advanced AI systems. We’re building smarter, more human AI with a diverse community in 100+ countries.
At Welo Data, Limitless AI. Limitless You. isn’t just a slogan—it’s our promise. We build smarter AI through the power of human contribution, offering limitless opportunities for our global community to grow, contribute, and work on their terms.