# Benetech - Making Graphs Accessible — Data

来源：https://www.kaggle.com/competitions/benetech-making-graphs-accessible/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

This competition's dataset comprises about 65,000 comprehensively-annotated scientific figures of four kinds: **bar graphs** (both *horizontal* and *vertical*), **dot plots**, **line graphs**, and **scatter plots**. While the majority of the figures are synthetic, we also include several thousand figures extracted from professionally-produced sources. Your task is to predict the data series depicted in the test set figures.

This is a [**Code Competition**](https://www.kaggle.com/competitions/benetech-making-graphs-accessible/overview/code-requirements), in which the actual test set is hidden. In this public version, we give some sample data drawn from the training set to help you author your solutions. When your submission is scored, this example test data will be replaced with the actual test set.

## File and Field Descriptions ##
- **train/annotations/** Collection of JSON image annotations describing the figures.
  - `source` Whether `generated` or `extracted`.
  - `chart-type` One of `dot`, `horizontal_bar`, `vertical_bar`, `line`, `scatter`.
  - `plot-bb` Bounding box of the plot within the figure, given by `height`, `width`, `x0`, and `y0`.
  - `text/id` Identifier for a text item within the figure.
  - `text/polygon` Region bounding the text item in the image.
  - `text/text` The text itself.
  - `text/role` The function of the text in the image, whether `chart_title`, `axis_title`, `tick_label`, etc.
  - `axes/{x|y}-axis/ticks/id` Identifier matching the tick to the associated text element `id`.
  - `axes/{x|y}-axis/ticks/tick_pt` Coordinates of each tick in the figure.
  - `axes/{x|y}-axis/tick-type` The graphical depiction of the tick element.
  - `axes/{x|y}-axis/values-type` The data type of the values represented by the tick element, whether `categorical` or `numerical`. This field determines how the predicted data series are scored. See the [**Evaluation**](https://www.kaggle.com/competitions/benetech-making-graphs-accessible/overview/evaluation) page for more information.
  - `visual-elements` Indicates part of figure representing the data series. Depends on the chart type.
  - `data-series/{x|y}` The x and y coordinates of values depicted in the figure. For the test set images, this is the target to be predicted.
- **train/images/** Collection of figures in JPG format to be used as training data.
- **test/images/** Collection of figures to be used as test data. Your task is to predict each corresponding data series for the figures in this folder. The public version of this folder contains only a few example images drawn from the training set.
- **sample_submission.csv** A submission file in the correct format. See the [**Evaluation**](https://www.kaggle.com/competitions/benetech-making-graphs-accessible/overview/evaluation) page for more details.

## Data Splits ##

The full test set comprises about 4,000 figures extracted from professionally-produced sources. The test set does not contain any generated figures.

The `extracted` figures in the training and public test sets are drawn from the same set of sources. The figures in the private test set are drawn from a distinct set of sources. You should ensure your solutions are robust to a variety of figure styles.

The distribution of chart types may not be identical in the public and private test sets.