Arbitrary-Shaped Text (ICDAR-2019 ArT)

Research Tasks

Scene Text Spotting

2019-05-29 (v. 1)

Contact author

Yipeng Sun

Baidu Inc, Beijing, China


+86 10 56082834

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.


The main objective of this task is to detect and recognize every text instance in the provided image in an end-to-end manner. We break this task down into two sub categories: T3.1 Latin script only text spotting; T3.2 Latin and Chinese scripts text spotting. 


For T3, we first evaluate the detection result by calculating its Intersection over Union (IoU) with the corresponding ground-truth. Detection regions with an IoU value higher than 0.5 will be matched with the recognition ground truth (i.e. the transcript ground truth of the particular text region). Meanwhile, in the case of multiple matches, we only consider the detection region with the highest IOU, the rest of the matches will be counted as False Positive. Then, we will evaluate the predicted transcription with both case-insensitive word accuracy H-mean and 1-N.E.D, while the Chinese text regions will be ignored in this evaluation for T3.1. We will calculate the results of T3.1 and T3.2 in both metrics (1-N.E.D and case-insensitive word accuracy), but the official ranking is based on the results of 1-N.E.D.

The expected output is a spatial location of every text instance at word-level for Latin scripts, and line-level for Chinese scripts together with the predicted word for each detection.


No comments on this dataset yet.

Add your comment

In order to comment on a dataset you need to be logged on
Register Now!


In order to rate this dataset you need to be logged on
Register Now!