liukuikun

I meet the same problem, have you solved it

![fsgnet]()

Before  After 

You can reference https://github.com/open-mmlab/mmocr/blob/main/mmocr/datasets/icdar_dataset.py and https://github.com/open-mmlab/mmocr/blob/main/mmocr/datasets/text_det_dataset.py which have difference annotation format. There are two ways for the custom annotation format: * Custom Dataset class and implement the method for parsing...

We have PR for ctw1500 at https://github.com/open-mmlab/mmocr/pull/1677

Thanks for your report! Maybe try to replace https://github.com/open-mmlab/mmocr/blob/0894178343af5589400ea145a861e709eef63071/mmocr/datasets/preparers/parsers/funsd_parser.py#L33 with `with open(file_path, 'r', encoding='utf-8') as f:`

1. maybe provide a sample cli for inference in https://mmocr.readthedocs.io/en/dev-1.x/get_started/quick_run.html#inference. 2. Maybe firstly introduce cli and infer.py args in the user guide and then task inferencer 3. Inferencer in user...

fix https://mmocr--1744.org.readthedocs.build/zh_CN/1744/get_started/install.html#id6

Current visualization can not show confidence scores. You can implement it at https://github.com/open-mmlab/mmocr/blob/689ecf0f5fca20975131e0c30552ebbfe526bddb/mmocr/visualization/textdet_visualizer.py#L66 by adding score parameter and drawing score text

Temporary inferring a custom image is not supported, because we are refactoring inference API. It will be implemented as soon as possible after https://github.com/open-mmlab/mmocr/pull/1608 "--work-dir" is used for saving logs...