youngmiao / crnn Goto Github PK
View Code? Open in Web Editor NEWBased on crnn add Chinese recognition
License: MIT License
Based on crnn add Chinese recognition
License: MIT License
你的输入图片尺寸高度为32,宽度时根据高度缩放的,是在脚本中将keep ratio设置为true吗
could you serve a copy of train.txt ,I can't train it for my own dataset,thanks a lot
first :
preds = preds.squeeze(2) -> preds = preds.squeeze(1)
second:
preds = preds.transpose(1, 0).contiguous().view(-1) -> preds = preds.transpose(-1, 0).contiguous().view(-1)
to change the two lines then get the result
我使用中文字符时,一个字符的len是3,当我使用100时,len为300,nclass为301,但是我看你使用了中文字符,类别数就是比字符数多1,所以我觉得有些疑问。还有一个问题就是我用合成的样本进行训练和测试,效果比较好,但是我用自然场景下扣出来的文本进行测试,如果只是效果不好那我觉得是合理的,但是我预测出来的标签序列有alphabet中不包含的字符,这让我觉得很奇怪。请问你有遇到过这个问题吗,或者你的看法是什么呢?
File "crnn_main.py", line 109, in
if len(model_dict[weig1]) == len(pre_trainmodel[weig1]) and len(model_dict[bias1]) == len(pre_trainmodel[bias1]):
KeyError: 'rnn.1.embedding.weight'
比如我有的训练图片有10个字符,而有的有13个字符,这种情况的训练集可以训练吗
@YoungMiao
I get an error in these lines. I think t7_data is an object, not a list
Lines 120 to 123 in e78bb7d
name = layer[0]
TypeError: 'NoneType' object is not subscriptable
麻烦问一下,我用多个gpu训练的loss感觉不对,与单gpu训练相同Loss值的模型,预测结果非常差,单gpu训练的模型预测是正确的。@ wulivicte,请问您遇到过这种问题吗,怎么解决的,非常感谢。
你好,我没有找到保存训练模型的代码,也不知道训练好的模型时保存在哪儿
max_iter = min(max_iter, len(data_loader))
accuracy = n_correct / float(max_iter * opt.batchSize)
请问这两句该怎么理解?
用synthdata-zh这个项目生成的图片,训练时一直有Corrupted image的错误
There is something wrong (dimenson mismatch) when fine tune with your code.
for k,v in model_dict.items(): if (k != weig1 or k != bias1): model_dict[k] = pre_trainmodel[k] crnn.load_state_dict(model_dict)
I fixed it by updating
if (k != weig1 or k != bias1):
to
if not (k == weig1 or k == bias1):
例如对于alphabet = 'ACIMRey万下依口哺摄次状璐癌草血运重' 的训练样本,是一个字一个字作为训练样本吗?还是词组性的,例如哺璐草?
我在执行过以下命令后:
git clone https://github.com/baidu-research/warp-ctc.git`
cd warp-ctc
mkdir build; cd build
cmake ..
make
在执行:
cd pytorch_binding
python setup.py install
发现pytorch_binding文件夹下没有setup.py文件
在编译安装warp_ctc的时候设置了CUDA_HOME,运行时依然提示no CUDA-capable device is detected
torch.cuda.is_available() 返回true
Hi, I am trying to add another language CRNN model in addition to English in my project [1] Do you have a pre trained model for chinese ?
Thanks,
您好!21类中英文大概是哪些类?
label imgdata = open("./train.txt")
是类似于 ./data/test00.jpg 1+1=2
./data/test01.jpg 1+2=3
.................
这样吗?
How to solve this problem?
你好。我想问一下。这个模型可以实现多卡训练嘛?
该怎么训练啊,不清楚训练的流程
为什么训练代码要用tensorflow模型?util文件夹下少了tflogger文件
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.