画像認識AIを活用した現場支援システムの開発
内装工事進捗と資機材位置を図面上に表示
時間外労働の削減が求められる中，ICT活用による現場施工管理の効率化・省力化は喫緊の課題である。施工管理における画像認識の有効性は高いと考えられ，とりわけAIへの期待は大きい。高層建物などの大規模現場では現場巡回に基づく工事進捗の確認作業に多大な時間と労力を要している。また，工事に使用する資機材は数千点に上ることがあり，現場内に点在する資機材の適切な管理が課題となっていた。そこで，現場業務の支援を行うことを目的に，現場巡回時に360度カメラで撮影した動画をAIで分析することで内装工事における壁・天井の施工状況および各種資機材の所在を自動認識し，図面上に表示するシステムを開発した。本報では，本システムの概要および現場での試行結果について示す。
キーワード：AI，画像認識，PDR，位置推定，ARマーカ
＊１ 技術センター 先進技術開発部 AI連携技術開発室
Development of a Support System for Building Sites Using AI for Image Recognition
Display of Interior Finishing Work Progress and Location of Various Materials and Equipment on Drawings
As reduction of overtime work is required, improvement of efficiency and labor-saving site construction management by the utilization of ICT are urgent issues. It is considered that the effectiveness of image recognition in construction management is high, and in particular there are great expectations for AI. On large-scale sites such as high-rise buildings, a lot of time and labor are required for confirmation of construction progress based on field patrols. Furthermore, the materials and equipment used for the construction sometimes reaches thousands of items, and appropriate management of the materials and equipment scattered throughout the site is a problem. Therefore, for the purpose of supporting work on site, a system which automatically recognizes the construction status of walls and ceilings in interior finishing work and the location of various materials and equipment was developed. In the system, video photographed by a 360-degree camera during a field patrol is analyzed using AI for image recognition, and the analysis results are displayed on a drawing. This paper presents an outline of the system and the field trial results.
Keywords: AI, image recognition, PDR, position estimation, AR marker
*1 AI Utilization Technology Development Section, Advanced Technology Development Department, Taisei Advanced Center of Technology