面向汽车智能化应用的《神经网络与深度学习》课程项目式教学改革探索

Exploration of Project-Based Teaching Reform in the “Neural Networks and Deep Learning” Course for Automotive Intelligence Applications

摘要: 随着汽车产业智能化升级，自动驾驶、智能座舱和车路协同等场景对兼具深度学习基础、工程实现能力与行业认知的复合型人才提出更高要求。针对“神经网络与深度学习”课程中存在的案例与产业场景衔接不够、理论教学与实践环节分离、课程考核偏重结果性等问题，结合新工科与成果导向教育理念，提出面向汽车智能化应用的项目式教学改革方案。以交通标志识别、车辆目标检测和驾驶员状态监测等任务为牵引，构建“基础理论–阶段实践–综合项目”的教学框架，设计“课前–课中–课后”实施流程，并围绕项目方案设计、阶段日志、成果答辩、个人贡献等环节构建多元过程性评价体系，可为人工智能类课程服务汽车产业需求的教学改革提供参考。

Abstract: With the rapid development of automotive intelligence, fields such as autonomous driving, intelligent cockpits, and vehicle-road coordination demand interdisciplinary talents with solid foundations in deep learning, engineering implementation skills, and industry awareness. To address the problems of insufficient alignment between course cases and industrial scenarios, separation between theory and practice, and result-oriented assessment in the “Neural Networks and Deep Learning” course, this paper proposes a project-based teaching reform scheme guided by emerging engineering education and outcome-based education. The reform uses progressive authentic tasks including traffic sign recognition, vehicle detection, and driver status monitoring, builds an integrated framework of “basic theory - phased practice - comprehensive project”, designs a three-stage process of “pre-class, in-class, and post-class”, and establishes a diversified process-oriented assessment system. In addition, the scheme emphasizes task-chain design, multi-module collaboration, and authentic industrial constraints, and can provide a reference for application-oriented reform of AI courses serving the automotive industry.