mxin262 / bridging-text-spotting Goto Github PK

请问训练的这个权重是什么呀，是检测的权重还是识别的权重呢

运行编译时 一直报错，尝试多种办法也无果，报错信息如下，能麻烦帮忙看看吗？

cc1plus: warning: command line option ‘-Wstrict-prototypes’ is valid for C/ObjC but not for C++
In file included from /home/[email protected]/Bridging-Text-Spotting/adet/layers/csrc/DeformAttn/ms_deform_attn_cpu.cpp:14:
/home/[email protected]/.conda/envs/Bridge/lib/python3.8/site-packages/torch/include/ATen/cuda/CUDAContext.h:5:10: fatal error: cuda_runtime_api.h: No such file or directory
5 | #include <cuda_runtime_api.h>
| ^~~~~~~~~~~~~~~~~~~~
compilation terminated.
error: command 'gcc' failed with exit status 1

论文中说不需要额外训练识别器，所以DiG模型的权重使用的是预训练权重而不是微调权重吗？
检测器使用的是微调权重？

I downloaded the annotation files for the tt dataset using the link you provided and successfully loaded the dataset and weights. However, I encountered this error during the forward process. I am not sure whether the issue is due to the code, my configuration, or a problem with a third-party library. I hope the author can provide some clarification. Thank you.

Hello;

Thank you for your great work. Can you guide me how can I train this model with custom dataset?

Thank you for your guidance

请教一下，用了这个命令做evaluation：python tools/train_net.py --config-file configs/Bridge/TotalText/R_50_poly.yaml --eval-only MODEL.WEIGHTS totaltext_testr.pth，但是出来报错是[COCOEvaluator] Did not receive valid predictions.
先前已经按照要求弄好了测试文件

Thanks for your outstanding work! I have noticed that the previous works use rotated datasets, like Total-Text. I want to know the difference between them and why the rotated images perform better than raw images.

Hello.
First of all, thanks for sharing nice works :>

I want to get a result of detection + recognition (text spotting).
But, it only show detection results without any text label.

I followed the rule that you mentioned like below
python demo/demo.py --config-file configs/Bridge/ICDAR15/R_50_poly.yaml --input input.png --output output.jpg --opts MODEL.WEIGHTS ./pretrained_models/Bridge_ic15.pth

Any reasons?

Thanks.

Dear author, I noticed that only the pretraining code is provided in the document, like
python tools/train_net.py --config-file configs/Bridge/TotalText/R_50_poly.yaml --num-gpus 4 MODEL.WEIGHTS totaltext_final.pth

Is there any code available for training from scratch?