chenhongruixuan / siamcrnn Goto Github PK

Hi, I would like to ask how the data in .pickle format like "train_sample_X.pickle" is obtained? Looking forward to your reply! Thanks!

In my opinion, the SiamCRNN is trained end-to-end to classify the central pixel of a patch as change or no change. The patches are of size 5x5 pixels, and the network attempts to classify the label of the central pixel based on its neighbourhood’s values.

When generating the training samples, do you specify the label (change or no change) of a pair of patches according whether the central pixel changed or not?

Hi,
A good work. In train.py file you load file train_sample_X.pickle , but I want to know how you write the images in this pickle file and where(in which file) I have to add this code.

请问作者，在数据集设置中您在各个pickle里是以什么形式存在，在之前的提问里看到lable是以（B,1）的形式存在，请问您能提供完整的训练集，以便我自己去训练一次模型吗，代码弄了很久还是没搞懂，希望您能帮助一下，不胜感激！！！

Hi, I want to know about the form of pikle format GF-2 dataset you use so that I can run siamCRNN on my own dataset. Or can the GF-2 dataset be provided?
Look forward to your reply. thank.

前辈您好！我又瞄上了您的这个代码哈哈哈哈~想请教您的label输入格式是什么呢,是图片吗，我应该把它和变化影像放在一个文件夹里吗？期待您的回复，谢谢！

Dear. author
Would you please share the heterogenous image with us to train a your SiamCRNN?
Thanks for your attention!!!

hi dear authors
your code have lots of problems. I runned it on colab but 'infer.py' and 'train.py' miss datasets. I guess you had data sets on your system so forget to upload it here. path also returns Null. you used load_data() func in 'infer_result()' of 'infer' but it doesn't have the input. unfortunately there is no 'train_sample_X and Y'.

For citing and fair comparisons, do you have plans to provide the pre-trained weights of the SiamCRNN as well as the dataset used in the paper?