tongkunguan/ccd

[ICCV2023] Self-supervised Character-to-Character Distillation for Text Recognition

Homepage

ccd's Issues

数据集的问题

您好，请问数据集在哪里能下载呢？感谢！

代码问题，请教

student = (
    student,
    SegHead(in_channels=config.model_seg_channel, mla_channels=128, mlahead_channels=64, num_classes=2),
    DINOHead(embed_dim, config.out_dim, use_bn=config.use_bn_in_head, norm_last_layer=config.norm_last_layer, ))
teacher = ABIDINOModel(
    teacher,
    None,
    DINOHead(embed_dim, config.out_dim, config.use_bn_in_head), )

请问这处代码是否有问题。

能否提供一个OCR_CC的百度网盘数据集链接？

感谢优秀的工作！ 能否提供一个OCR_CC的百度网盘数据集链接呢？由于数据集过大，DiG的原始数据链接难以下载。

NameError

Dino/dataset/dataset.py
Line 16 name 'Dataset' is not defined !
Line 20 name 'PathOrStr' is not defined !
Line 40 name 'Path' is not defined !

环境问题

您好，请问这些包yaml opencv-python Pillow LMDB nltk six natsort scipy sklearn scikit-image matplotlib editdistance tqdm大致的版本是多少呀，直接pip会出现各种版本问题

训练时长问题

请问您是用4张3090跑的预训练吗？跑了多久啊？

Runtime error in dino_vision.py file: shape '[16, 8, 32, 512]' is invalid for input of size 131072

Hello,

I currently training your model (vit_base) on my own dataset, but this issue came up:
RuntimeError: shape '[16, 8, 32, 512]' is invalid for input of size 131072
This error came from line 56 in Dino/model/dino_vision.py

The backbone_out shape is torch.Size([16, 16, 512])

region_f = backbone_out.reshape(N, 8, 32, E).permute(0, 3, 1, 2)

Can you please explain why 8, 32 in this line?

This issue came up because I changed the patch_size=16 and use_fp16=True in the config file, but if I still remain patch_size=4 then it can normally run.

I run with batch_size_per_gpu=16

请问table中的数据集STD是单纯指使用合成数据集预训练吗？还是使用STD+URD？

还是我应该理解成URD数据集是单纯用来做预训练的。data的：STD和ARD用来做微调，十分感谢~

Is it possible to run a single image Demo?

Hi, thank for your great work!

I have read the paper and now I am trying to test a single image OCR. Is it possible to run a single image test?