Image classification with 96% accuracy is achieved by ReVision. They used 2500 chart images that were collected from the Internet and were divided into 10 groups, which are called area, bar, Pareto, and pie charts, curve, radar, and scatter plots, tables, and Venn diagrams.
, aims at classifying and extracting data from charts. To get what a chart means, many people have done some amazing jobs related to research areas. The information that we get from the chart can provide data support for the specific applications of ubiquitous information. So, to recognize a chart, and get the information from that, is very meaningful.
The box chart of stock price cannot be ignored during such a task. For example, we want a deep learning algorithm to help us predict the future price of some stocks. To automatically analyze a chart, then getting information (usually original table data) from it can bring us huge benefits. Though it is a very easy task for people to read and understand the chart, a very hard task for a computer. Ī chart is always created from a table once a chart has been painted to image, the original table information is lost. It is very intuitive, people can read the data of a chart without effort, even more, people can read some other information from a chart at a glance without calculation, such as the longest and shortest ones, the trend of the data. The chart has been widely used for data visualization. Ubiquitous information is mainly divided into charts, texts, images, audios, and videos according to the form of expression. People spread and communicate all aspects of content through the ubiquitous network, which is called the “ubiquitous network.” Through the ubiquitous network, anyone can obtain any information they need at any place, time, and location, which is called ubiquitous information. A vast amount of data are produced every day. The era in which we have gone through is called the age of the data. Finally, we utilize some traditional geometric methods to obtain an original table of a chart, so we can redraw it. Besides, we also introduce two datasets, UCCD and UCID, to train deep models to classify and recognize charts. When recognizing the structure and content of charts, we use different deep frameworks to extract key points based on different types of charts. When classifying charts, we choose ResNet-50. This study proposes a three-stage method to chart recognition: analyze the classification of charts, analyze the structure of charts, and analyze the content of charts. Due to there are so many different kinds and different styles of charts, it is not an easy task for a computer to recognize a chart, as well as to redraw the chart or redesign it. Chart is one kind of ubiquitous information, which is widely utilized and easy for people to understand.