MaskTextSpotter

MaskTextSpotter copied to clipboard

test issue with TEST.IMS_PER_BATCH greater than 1

3

Hi all, I'm running text spotting on batches of images, with TEST.IMS_PER_BATCH = 16. Some error raises from function process_char_mask in text_inference.py, that the length of boxes doesn't match char_masks.shape[0]...

sshl

Evaluation for Total-Text dataset

1

Thanks for your excellent work! Any plans for releasing codes of Total-Text evaluation?

TriplePool

And how can I train my own recognition only model?

JimmyJuan

if 'predictions' is None, we can return (None, None) https://github.com/MhLiao/MaskTextSpotter/blob/afe22795833dd93bc0b208de06b768ff1144eba9/tools/demo.py#L89

vishaljainai

In TextDemo.compute_predictions, the method appends `polygons` to `result_polygons` before checking confidence score, resulting in occational orphaned polygons with no text or score and mismatched polygon and text in the visualisation.

kwsp

whai362

@MhLiao If I want to use it to detect just numbers and '.', what should I do? Thanks

arthurchanhokman

how to train from the begin

6

The pretrain code loaded the trained model to get the model, how to train from the begin without downloaded model? the loss seems to be very small.

ChChwang

Evaluation code

2

when will the evaluation code be released?

ChChwang

Thanks to the author for sharing the code, I would like to ask you, for example, if I add 50 Chinese characters, what should I modify?

ustczhouyu