mmocr/configs/textrecog/seg
lizz a48399aebe format again
Signed-off-by: lizz <lizz@sensetime.com>
2021-04-05 16:39:31 +08:00
..
README.md format again 2021-04-05 16:39:31 +08:00
seg_r31_1by16_fpnocr_academic.py add sar, seg and other components 2021-04-02 23:54:57 +08:00
seg_r31_1by16_fpnocr_toy_dataset.py add sar, seg and other components 2021-04-02 23:54:57 +08:00

README.md

SegOCR Simple Baseline.

Introduction

[ALGORITHM]

@unpublished{key,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  note={Unpublished Manuscript},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

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

Notes:

  • 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.