News - ICDAR2017 Robust Reading Challenge on end-to-end recognition on the Google FSNS dataset

ST-VQA Training Set Updated

The training set of the ST-VQA challenge has been updated. Apart from filtering noisy data, we have made slight changes in the JSON structure to make it more readable. Most notably, in the training set JSON file for Task 1, the key "distractors" has been renamed to "dictionary".