Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

有几个现在无法处理的问题需要请教帮忙....... #25

Open
IALin opened this issue Oct 4, 2018 · 9 comments
Open

有几个现在无法处理的问题需要请教帮忙....... #25

IALin opened this issue Oct 4, 2018 · 9 comments

Comments

@IALin
Copy link

IALin commented Oct 4, 2018

我想要在不训练,使用现有模型的情况下实现图片内文字检测的功能,阅读文档后还有以下的问题:
1、直接使用已有模型,是否需要从tutorial2的部分开始进行呢?还是直接从tutorial3中间开始呢?
2、如果需要下载一部分的图片文件来避免程序报错,是否应该放置在../data/image/下的test和trainval文件夹下?
3、我想要采用google inception模型,下载之后解压,得到的模型文件应该放到哪里?
4、如果想要实现自己的图片检测,图片路径应该在哪里?

@yuantailing
Copy link
Owner

  1. From tutorial 3.
  2. Yes.
  3. When you run test script, notice which file is missing, then you should download model to there.
  4. Change detection/prepare_test_data.py to read your own data.

@IALin
Copy link
Author

IALin commented Oct 4, 2018

谢谢,后面还有问题,还希望能够帮忙。百忙中打扰您了。

@IALin
Copy link
Author

IALin commented Oct 5, 2018

您好,还有几个问题,我现在依旧无法处理,再次打扰您了。问题如下:
1.看了一下您的论文Chinese Text in the Wild
在4.2部分的第一段提到
4.2. Detection
Given an image, the goal of the character detection task
is to detect the bounding boxes of all character instances
and also recognize each character instance, i.e., predict its
character category.
里面提到,detection的作用除了标出汉字的边界框之外,还有识别每一个汉字类别的功能。请问这个predict character category是直接得到最终的文字结果吗?

2.之后提到
.Following the recognition task (Section 4.1), we also
limit the number of categories to 1001,

3.请问classification和detection这两个部分是否存在关联呢?

4.我们已经借助现有的模型生成了eval_inception_v4.pkl 、 predictions_inception_v4.jsonl两个文件,并且打开jsonl可以看到一些结果。但是detection部分似乎并没有用到classification的模型和结果。

@yuantailing
Copy link
Owner

1 Yes, see last figure.
3 The relationship is explained in paper.
4 Detection and classification are two different tasks.

@IALin
Copy link
Author

IALin commented Oct 5, 2018

谢谢您的帮助,真的帮了我许多。

@IALin
Copy link
Author

IALin commented Oct 6, 2018

One more question!!!!! I'm Sorry!

在使用自己的图片进行测试的时候,需要将图片按照规定的格式命名吗?(根据远近、是否遮挡等命名成测试集中图片的样子)还是可以直接命名为test1.jpg等等呢?

@yuantailing
Copy link
Owner

It reads image file from "file_name" key: https://github.com/yuantailing/ctw-baseline/blob/master/detection/prepare_test_data.py#L49

Nowhere else use "file_name" key, so there is no limit on "file_name".

@IALin
Copy link
Author

IALin commented Oct 6, 2018

您好,还有一些小问题:
1.我修改了detection路径下的settings.py文件,将TEST_IMAGE_DIR的指向改成了另外一个文件夹test2,并且解压了原测试集的一部分图片进去。接着我尝试运行python3 prepare_test_data.py,提示AttributeError: 'NoneType' object has no attribute 'shape',这应该是找不到图片的意思,但是我们确实已经放了图片了,请问这是怎么回事呢?(和软链接有关吗?)

@yuantailing
Copy link
Owner

yuantailing commented Oct 6, 2018

You should change the code to adapt to your data by yourself.

I can only help you to reproduce our experiment results, that's all.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants