In order to solve the problem of real-time and on-site automatic 3D modeling of distribution network equipment, a 3D modeling method based on photos of these distribution network equipment taken by cell phone was proposed. Key technologies of the method include two aspects. Firstly, in order to realize the size detection of distribution network equipment based on their photos, a Unet enhanced algorithm was proposed to extract the outline of equipment from their photos, then an enhanced line segment detector(LSD) method was proposed to obtain outline size of the equipment, and an improved K-nearest neighbor(KNN) algorithm was proposed to search the equipment specification size. Secondly, a data-driven fast and automatic parameteric 3D modeling method was proposed, where the equipment specifications and sizes were stored in MySQL database as driving parameters, and the MaxScript script program of 3ds Max was linked and driven by MySQL database to automatically complete the 3D modeling of distribution network equipment in real time. Case study of verification was carried on transformers, and the results showed that the proposed method effectively overcomes the difficulties in on-site manual size measurement and inefficient modeling of distribution network equipment, thus provides strong technical support for upgrading the 3D visual management of distribution network.