Element Layout Prediction with Sequential Operation Data

Creating the regular layout of elements is a fundamental task in many scenarios.This task is often te-dious since the user has to perform a lot of repeated operations.To address this problem,a layout prediction method is presented to assist the user for the regular layout creation task.30 groups of flow chart template are collected as the basic data set from SmartDraw and data enhancement operation is applied.These data are en-coded in the form of image and property list and are used to train a neural network.The trained network is able to learn the layout patterns and perform the layout prediction.The prediction includes multiple layout candidates in different scales and reaches an acceptable accuracy.Proposed method is tested with a user study including both simple and complex tasks.The evaluation is based on the quantitative analysis of the performance,effectiveness,and stability.The results show that,compared with traditional tools,proposed method is able to notably reduce the number of element positioning operations and the task completion time,and therefore improves the efficiency of layout creation.