Giter VIP home page Giter VIP logo

Comments (16)

zhijiezhong avatar zhijiezhong commented on July 19, 2024

你好,报错可能是pycharm自身原因,我的就没有报错,那里import的是这个目录下的文件,有这些文件肯定可以运行

from crnn.

nissansz avatar nissansz commented on July 19, 2024

哦。我mark as source 好像也没用。
690张图,训练2轮,测试结果全是失败,是正常的吗?要怎样才能正确识别?

C:\Users\Ni\AppData\Local\Programs\Python\Python38\python.exe F:/pycharm2020.2/crnn-master/test.py
load weights/chinese_2_82.80496325595773_0.0.pt model success!
device: cpu
test_batch_loss=84.44302188648896, test_batch_acc=0.0
test_loss=84.44302188648896, test_acc=0.0
但是先不说妈妈 ------------> ❌❌❌
我绝对不想成为爸爸那样的人 ------------> ❌❌❌
我爸爸逃避着自己的问题 ------------> ❌❌❌
只替周围的人操心 ------------> ❌❌❌
有没有我能做的事呢 ------------> ❌❌❌
我想我一定无法成为那样的人 ------------> ❌❌❌
但是在这个世界上 ------------> ❌❌❌
有很多逃避自己问题的人 ------------> ❌❌❌
这样的人很轻松 ------------> ❌❌❌

from crnn.

zhijiezhong avatar zhijiezhong commented on July 19, 2024

正常,我训练了一天准确率才是96,你现在训练太少了,在训练集的准确率还是0。我训练用的是100万的数据集,训练用lr=0.0001训练的3个epoch,再lr=0.00001训练2个epoch

from crnn.

nissansz avatar nissansz commented on July 19, 2024

lr=0.0001 在哪里设置?
您训练的模型可以分享吗?我想测试看看准确率。

from crnn.

zhijiezhong avatar zhijiezhong commented on July 19, 2024

模型:https://cloud.189.cn/web/share?code=U7NV3iNbmYvi
数据集:https://cloud.189.cn/web/share?code=eyuiiyjyEFva(访问码:4giq)

from crnn.

nissansz avatar nissansz commented on July 19, 2024

谢谢。自己准备的训练用的图片,要不要先用ctpn的定位把它裁边裁成只剩文字的?每张图的尺寸要统一吗?

from crnn.

zhijiezhong avatar zhijiezhong commented on July 19, 2024

image
学习率这里设置

from crnn.

zhijiezhong avatar zhijiezhong commented on July 19, 2024

需要裁剪成只有文字,但不需要统一尺寸,我会在读取数据集时处理成32*1000的大小。这里的1000你可以改,改源代码

from crnn.

nissansz avatar nissansz commented on July 19, 2024

下载了你那个chinese_4_0.4110666987350187_0.95999.pt
我用我自己的图片测试为什么报错?必须要用训练的图片才能有结果?

C:\Users\Ni\AppData\Local\Programs\Python\Python38\python.exe F:/pycharm2020.2/crnn-master/test.py
Traceback (most recent call last):
File "F:/pycharm2020.2/crnn-master/test.py", line 91, in
main(opt)
File "F:/pycharm2020.2/crnn-master/test.py", line 62, in main
crnn.load_state_dict(torch.load(opt.weights, map_location=torch.device('cpu')))
File "C:\Users\Ni\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\modules\module.py", line 1482, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for CRNN:
size mismatch for rnn.1.embedding.weight: copying a param with shape torch.Size([2569, 512]) from checkpoint, the shape in current model is torch.Size([776, 512]).
size mismatch for rnn.1.embedding.bias: copying a param with shape torch.Size([2569]) from checkpoint, the shape in current model is torch.Size([776]).

from crnn.

zhijiezhong avatar zhijiezhong commented on July 19, 2024

你发一张图片让我测试一下

from crnn.

nissansz avatar nissansz commented on July 19, 2024

识别时是一定要用二值化图?还是场景文字也可以?
下面有4张图。2张中文,2张日文。
识别时,也要用ctpn裁剪成只剩文字区域?然后二值化,再进行识别?
0_00_21_016__0_00_22_351_0000000000000000000000001
0_00_21_016__0_00_22_351_0000000000000000000000001

0_00_21_016__0_00_22_351_0000000000000000000000001

0_00_21_016__0_00_22_351_0000000000000000000000001

from crnn.

zhijiezhong avatar zhijiezhong commented on July 19, 2024

我在读取图片时就二值化了

from crnn.

nissansz avatar nissansz commented on July 19, 2024

我在读取图片时就二值化了

但是如果我给的是白底黑字的图,没必要二值化吧?

from crnn.

nissansz avatar nissansz commented on July 19, 2024

chinese_7_18.998092761142647_0.0

18.998092761142647 是代表什么?

如果我按默认设置先训练10epochs
然后可以用weights, 继续接着最后生成的pt文件,继续训练?

from crnn.

zhijiezhong avatar zhijiezhong commented on July 19, 2024

chinese_7_18.998092761142647_0.0代表的意思是 chinese_epoch_loss_acc。可以加载保存好的权重继续训练

from crnn.

zhijiezhong avatar zhijiezhong commented on July 19, 2024

你好,加你微信了,回复一下

from crnn.

Related Issues (1)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.