s3nh_

4 ms·

1. 1.▲
   Show HN: Text Detection Tool(github.com)

   1 points by s3nh_ 7y ago|0 comments
2. 2.▲
   by s3nh_ 7y ago

   Hi, thanks for feedback! I'll add more general information. In my opinion theres a lot to do in complex document classification, I'll try to add some demo to make things more intuitive. thanks!
3. 3.▲
   by s3nh_ 7y ago

   Looks really nice! Love to see it next to some Rust-like or Golang=like backend. What do you think?
4. 4.▲
   by s3nh_ 7y ago

   the hardest part in training model in foreign languages is to get correctly labeled dataset. I worked with pretrain model on Polish language documents and based on this experience it is relatively good if you are using some text similarit
5. 5.▲
   by s3nh_ 7y ago

   In the example, text is not sorted by it's corrdinates but by appearence of boxes in first network. It is visible in more complex documents, that crnn network did not create boxes in descending order (word-by-word). also, good point ab
6. 6.▲
   Show HN: Pytorch Text Recognition Tool(github.com)

   41 points by s3nh_ 7y ago|8 comments