Open Asad-Raza opened 1 year ago
Have you tried rotating the image 90 degrees and redo a detection again, and obtain the union of bounding boxes in both original and rotated orientations?
No , I have not tried rotating the image. I will try this. Thank you for the suggestion
I have the same trouble, I'm trying to solve this, but it's to difficult to me... I think convolutional neural networks, its the best option to solve this problem...training word by word.
Hello
CRAFT is doing great in detection of horizontal text but it is not able to detect the text in vertical orientation properly. Can you provide guidance/suggestions on how to achieve detection of vertical text with CRAFT? Will training with images with vertical orientation suffice?
Thanks