Skip to content

hanson-young/ocr-table-ssd

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

23 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Improved-JNetV3 SSD in Pytorch

This repo is forked from pytorch-ssd.

  • We use MobileNetV2 as basenet
  • Change image input size form 300x300 to 768x768, it is propitious to the details of table
  • We use improved JNetV3 segmentation network, treated as a Auxiliary Supervisory Signal.It can accelerate the training process(get same loss : 1 night->20 min) and get a better accuracy
  • Change anchor for long width with short height table ceil

Improved-JNetV3 SSD architecture

Example of Improved-JNetV3 SSD architecture

original MobileNetV2 SSD

Example of MobileNetV2 SSD

Improved-JNetV3 SSD

Example of Improved-JNetV3 SSD

Improved-JNetV3 SSD Heatmap

Example of Improved-JNetV3 SSD Heatmap

Dependencies

  1. Python 3.6+
  2. OpenCV
  3. Pytorch 0.4+
  4. Caffe2
  5. Pandas
  6. Boto3 if you want to train models on the Google OpenImages Dataset.

Train

train_ssd.py #train file

Test

eval_ssd.py #test file

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages