History

liukuikun e1e26d3f74 [Enchance] add codespell ignore and use mdformat (#1022 ) * update * update contributing * update ci * fix md * update pre-commit hook * update mdformat Co-authored-by: gaotongxiao <gaotongxiao@gmail.com>		2022-06-09 14:58:44 +08:00
..
README.md	[Enchance] add codespell ignore and use mdformat (#1022 )	2022-06-09 14:58:44 +08:00
metafile.yml	unify task names in metafiles (#548 )	2021-11-03 15:03:37 +08:00
seg_r31_1by16_fpnocr_academic.py	[Refactor] refactor textrecog config structure (#617 )	2021-11-25 16:27:45 +08:00
seg_r31_1by16_fpnocr_toy_dataset.py	[Refactor] refactor textrecog config structure (#617 )	2021-11-25 16:27:45 +08:00

README.md

SegOCR

Abstract

Just a simple Seg-based baseline for text recognition tasks.

Dataset

Train Dataset

trainset	instance_num	repeat_num	source
SynthText	7266686	1	synth

Test Dataset

testset	instance_num	type
IIIT5K	3000	regular
SVT	647	regular
IC13	1015	regular
CT80	288	irregular

Results and Models

Backbone	Neck	Head			Regular Text			Irregular Text	download
				IIIT5K	SVT	IC13		CT80
R31-1/16	FPNOCR	1x		90.9	81.8	90.7		80.9	model \| log


-   `R31-1/16` means the size (both height and width ) of feature from backbone is 1/16 of input image.
-   `1x` means the size (both height and width) of feature from head is the same with input image.

Citation

@unpublished{key,
  title={SegOCR Simple Baseline.},
  author={},
  note={Unpublished Manuscript},
  year={2021}
}