vbhavank/Siamese-neural-network-for-change-detection

detail question

sunbin1205 opened this issue · 17 comments

sunbin1205 commented

ImportError: No module named Concatenate

vbhavank commented

We are currently working on testing the code and we do realize there are many installation issue. Will post a detailed instructions to run the code soon. Thank you for your patience.

vbhavank commented

@sunbin1205 Meanwhile which version of tensorflow and Keras are you using?

vbhavank commented

@sunbin1205 The concatenate script was uploaded. Thank you for your patience.

sunbin1205 commented

Glad to receive your reply!
I was so busy recently that I didn't reply in time. I'm sorry for that.
I have a few questions about the code you share:

  1. The input. txt file in the file contains the training data, right?
  2. Is the code disclosed now contains all the code files needed to complete the change detection of satellite images?
  3. Can you give me some test samples and test results for me to observe?
  4. I am now working on a project of change detection of buildings in satellite images. Do you have any advice?
    I'm sure your advice will be of great help to me!
    Looking forward to your reply, thank you!
vbhavank commented

@sunbin1205

  1. Yes, the input.txt file contains the path to the training data.
  2. Yes, it should contain all the required scripts, Please get back if you think there are additional scripts missing.
  3. Unfortunately, the data we used for our model are classified and i am not permitted to share and data. I am sorry.
  4. I expect the buildings to work the same as my original problem statement while creating the code. You could try if the performance increases by just using 1 channel(gray scale) as i don't expect the RGB channels to hold significant information in the case of buildings.

I apologize for the delay, i have been busy working on my thesis.
Thank you for your patience.

vbhavank commented

@zhangyuxin0 That's great! No i am not Chinese but i can use an online translator to reply to your queries.

Looking forward to your questions.

ormosia6 commented

I am postgraduate in China.Can we add Wechat to talk about this question?My Wechat number is 15700708104.

vbhavank commented

@ormosia6 I do not use Wechat but you can reach out to me via email at bxv7657 at rit dot edu.

Will be happy to answer any questions you have.

luojiahongye commented

Hi, can you share the files 'model.json' and 'model6.h5'. You said the training data are classified but I think the model is not. Thanks!

vbhavank commented

Anything that was produced from the data cannot be shared.

Rikaaaaaa95 commented

I've been learning about it recently. Do I learn from each other? Are you Chinese?

你好，能否留一下QQ呢

Hao-Nan commented

I've been learning about it recently. Do I learn from each other? Are you Chinese?

请问这个代码中的数据集从哪儿可以下载呢？非常感谢！

xd-heling commented

Excuse me, what are the functions of model.json and model6.h5? ? ? ? ? ? ? ? ? ? ?

zyp19 commented

Excuse me, what are the functions of model.json and model6.h5? ? ? ? ? ? ? ? ? ? ?

Did you run it successfully?What files are "modelh6.h5" and "model.json"?

Thank you very much!

xd-heling commented
zyp19 commented

The "modelh6.h5" and "model.json" are defined in  the "Train.py" file.   "model.json"  is the decision layer structure and "modelh6.h5" is the weight file. . This is the  structure of the decision layer.
…
------------------ 原始邮件 ------------------ 发件人: "zyp19"<notifications@github.com>; 发送时间: 2020年4月3日(星期五) 晚上8:30 收件人: "vbhavank/Siamese-neural-network-for-change-detection"<Siamese-neural-network-for-change-detection@noreply.github.com>; 抄送: "えいえんの ぜろ"<876027023@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [vbhavank/Siamese-neural-network-for-change-detection] detail question (#1) Excuse me, what are the functions of model.json and model6.h5? ? ? ? ? ? ? ? ? ? ? Did you run it successfully?What files are "modelh6.h5" and "model.json"? Thank you very much! — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

Thanks for your anwering. By the way, is the weight file the pre-training weight for vgg16?And are there two files defined here?If it's convenient for you, could you add my WeChat -- "ZoeZoeZzzzzz".

model_json = model.to_json()
with open("model.json", "w") as json_file:
      json_file.write(model_json)
# serialize weights to HDF5
model.save_weights("model6.h5")
print("Saved model to disk")
zyp19 commented

The "modelh6.h5" and "model.json" are defined in  the "Train.py" file.   "model.json"  is the decision layer structure and "modelh6.h5" is the weight file. . This is the  structure of the decision layer.
…
------------------ 原始邮件 ------------------ 发件人: "zyp19"<notifications@github.com>; 发送时间: 2020年4月3日(星期五) 晚上8:30 收件人: "vbhavank/Siamese-neural-network-for-change-detection"<Siamese-neural-network-for-change-detection@noreply.github.com>; 抄送: "えいえんの ぜろ"<876027023@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [vbhavank/Siamese-neural-network-for-change-detection] detail question (#1) Excuse me, what are the functions of model.json and model6.h5? ? ? ? ? ? ? ? ? ? ? Did you run it successfully?What files are "modelh6.h5" and "model.json"? Thank you very much! — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

Do I understand that correctly?Are these two files automatically generated by the model and not imported from somewhere else?