产教融合驱动下人工智能赋能机器视觉实验课程改革与实践

Reform and Practice of an AI-Empowered Machine Vision Experimental Course Driven by Industry-Education Integration

摘要: 针对传统《机器视觉》实验课程验证性内容偏多、工业情境融入不足及工程能力评价较为主观等问题，以矿冶企业浮选泡沫智能感知任务为载体，构建产教融合驱动的人工智能赋能实验教学模式。课程基于项目式学习、基于问题的学习与情境学习理念，按照“工业问题导入–数据建构–模型实践–工程优化–评价反馈”组织图像采集、数据标注、实例分割、模型评价及分层拓展。研究采用能力评价量表对改革班和传统班的项目成果进行匿名双评，并结合学生问卷和多主体反馈分析教学效果。结果表明，改革班综合能力评价得分高于传统班，优势主要体现在数据建构与质量控制、结果分析与指标解释以及工程适配与方案决策等方面。本研究为地方行业特色高校将真实企业任务转化为可实施、可评价的人工智能实验项目提供参考。

Abstract: To address the excessive emphasis on verification-oriented activities, insufficient industrial context, and subjective assessment of engineering competence in traditional “Machine Vision” laboratory courses, this study develops an AI-empowered teaching model driven by industry-education integration, using intelligent perception of flotation froth as the project carrier. Drawing on project-based learning, problem-based learning, and situated learning, the course follows the chain of “industrial problem introduction-data construction-model practice-engineering optimization-evaluation and feedback” and integrates image acquisition, data annotation, instance segmentation, model evaluation, and tiered extension tasks. A competency-based rubric is used for anonymous dual-rater assessment of project artifacts from a reform class and a traditional class, supplemented by student questionnaires and multi-stakeholder feedback. The results showed that the reform class achieved a higher overall competency assessment score than the traditional class, with the main advantages observed in data construction and quality control, result analysis and metric interpretation, and engineering adaptation and decision-making. The study provides a replicable reference for industry-oriented universities to transform authentic enterprise tasks into assessable AI laboratory projects.