基于图像分割与轨迹追踪的室内饰面施工进度智能评估方法

施工进度是工程项目管理的关键组成部分，传统的进度管理多依靠人工巡检，耗时费力且无法保障进度评估的时效性。为了实现施工进度的自动化、高效化监管，针对室内施工湿作业铺贴场景，提出一套智能化进度追踪与评估框架，基于改进的Mask R-CNN深度学习框架，自动提取室内墙面和地面的瓷砖铺贴进度，基于相机轨迹追踪算法，将整个施工层的大范围施工进度识别结果映射至BIM可视化模型，为施工数字孪生奠定基础。在上海市某高层建筑项目中进行实例研究，实现了施工图像的高精度分割，验证了研究框架的可行性。该框架不仅适用于室内瓷砖铺贴，也适用于抹灰、刷涂等室内装修工程。

Intelligent evaluation method of indoor finishing construction progress based on image segmentation and positional tracking

Construction progress is a key part of project management. Traditional progress management mostly relies on manual inspection, which is time-consuming and can not guarantee the progress evaluation. In order to automatically and efficiently monitor construction progress, this article proposed an intelligent framework for indoor construction progress evaluation. This framework can automatically extract the tile laying progress of indoor wall and ground based on the improved Mask R-CNN, and map the progress results to the BIM for visualization using camera tracking algorithm. The framework was successfully applied in a building project in Shanghai with a high precision of image segmentation, verifying the feasibility of the presented framework. This study provides theoretical and practical reference for the automated progress tracking and unmanned construction progress supervision of indoor continuous space.