-
I use four GTX 1080 GPUs to train ABCNet with the default configs of syntext dataset. At the beginning of training, it goes normal situation, but the CUDA memory becomes larger as the training goes on…
-
Thanks for your work!
In the paper, the Total-Text has GT of 10 annotated points and CTW1500 has GT of 14 annotated points. But in my custom datasets, each curve text 's number of annotated points ar…
-
Thanks for your great work and sharing the code.
I wonder what is the number of point to annotate the curved text area.
-
Hi, thanks for your great work!
But when I run the ABCNet demo, it occurred an error.
Just for a simple test, I use the my own dataset, following your demo command:
```
python demo/demo.py \
…
-
I would like to annotate images with Russian text.
I don't understand how to create all the annotation.
For example, for TotalText data should be seen as follows:
`{"image_id": 8,
"bbox": [406.0,…
-
My Question is form into many sets, here are they,
# Section 1 | About Annotation Set
- Total-Text contains `.txt` formatted annotation. Like this:
```
x: [[214 280 362 349 284 231]], y: […
-
Training with custom datasets, programing pause more than 10m with fellowing log, haven't print training meg. I have prepare dataset via example script and checked carefully with outputed dataset. An…
-
Thank you for your work!
I have tried to quantize ABCNet but have been unsuccessful due to the below reasons:
- Some code not supported by torch.jit
- Some errors by Quant/DeQuant
Could you p…
-
I am training ABCNet on my own data having 300 training images.
I prepared my data using the [scripts](https://universityofadelaide.box.com/s/fo7odnmqe370btm7sdotqve1c0zsu8p3) you have provided
I am…
-
hey, in the abcnet paper, you showed evaluation on test set using `F-measure`. But I'm confused, how you measure this for evaluating the text recognition. What's the intuition behind. How is it doing …