Abstract: This paper proposes a method for automatically recovering data from chart images. In particular we focus on the task of estimating line charts, as the most common chart type, in a …

We estimate pixel-wise probabilities of different drawing styles using a segmentation network, build a graph and apply line tracing using linear programming. In order to obtain precise …

Authors: Hajime Kato (Rakuten Institute of Technology); Mitsuru Nakazawa (Rakuten Institute of Technology, Rakuten Group, Inc.)*; Hsuan-Kung Yang (National T...

Jan 1, 2022 · Examples from the SYN dataset, containing 40,000 automatically generated chart images with ground truth annotation. Statistical error analysis. We analyzed the causes of …

This paper proposes a method for automatically recovering data from chart images. In particular we focus on the task of estimating line charts, as the most common chart type, in a fully …

This paper proposed a novel framework to automatically re- cover data from line chart images using a single segmenta- tion network to extract line type probabilities. Line fitting was …

Aug 19, 2023 · In this work, we propose a robust method for extracting data from line charts by formulating it as an instance segmentation problem. Our main contributions can be …

Line Chart Data Extraction is a natural extension of Op-tical Character Recognition where the objective is to re-cover the underlying numerical information a chart image represents. Some …

This paper proposes a method for automatically recovering data from chart images. In particular we focus on the task of estimating line charts, as the most common chart type, in a fully …

This repository contains scripts to synthesize vertical bar charts, horizontal bar charts, pie charts and line charts as pairs of images (PNG) and their annotations (JSON), which are used for the …