Changes to upgrade libs.

[kannangce]

Fixes to have TF2.x usable

[h476564406]

@tonybeltramelli hello, this project is very interesting, thank you so much.

However, I have encountered a problem that is very puzzling, if you could give some advice and insights, I would greatly appreciate it.

The last step， I run
./generate.py ../bin pix2code /ossfs/node_41166019/workspace/pix2code/datasets/web/eval_set ../code greedy

The result seems overfitted, all gui generated from eval_set to code, is same DSL . I try to use other images , always generate the same DSL as well。

The same DSL content:

header{
btn-inactive,btn-active,btn-inactive,btn-inactive,btn-inactive
}
row{
single{
small-title,text,btn-orange
}
}
row{
double{
small-title,text,btn-orange
}
double{
small-title,text,btn-orange
}
}
row{
quadruple{
small-title,text,btn-red
}
quadruple{
small-title,text,btn-orange
}
quadruple{
small-title,text,btn-orange
}
quadruple{
small-title,text,btn-orange
}
}
row{
double{
small-title,text,btn-red
}
double{
small-title,text,btn-orange
}
}

@kannangce Do you have encountered a problem seems the result overfiltted? Thank you so much if you can give me some guides.

[integral-llc]

as of Apr 2, 2024, I found these libraries to be useable.

Keras==3.0.0
numpy==1.23.5
opencv-python==4.9.0.80
h5py==3.10.0
tensorflow==2.16.1

[h476564406]

已经收到。

[jtshushu]

(py36) E:\koutu\pix2code-xin\pix2code-master\model>python ./build_datasets.py ../datasets/ios/all_data
Splitting datasets, training samples: 1499.142857142857, evaluation samples: 249.85714285714286
Traceback (most recent call last):
File "./build_datasets.py", line 79, in
assert len(eval_set) == evaluation_samples_number
AssertionError

Why does such a problem occur and needs a solution?

[ChaymaJaffel]

@tonybeltramelli hello, this project is very interesting, thank you so much.

However, I have encountered a problem that is very puzzling, if you could give some advice and insights, I would greatly appreciate it.

The last step， I run ./generate.py ../bin pix2code /ossfs/node_41166019/workspace/pix2code/datasets/web/eval_set ../code greedy

The result seems overfitted, all gui generated from eval_set to code, is same DSL . I try to use other images , always generate the same DSL as well。

The same DSL content:

header{
btn-inactive,btn-active,btn-inactive,btn-inactive,btn-inactive
}
row{
single{
small-title,text,btn-orange
}
}
row{
double{
small-title,text,btn-orange
}
double{
small-title,text,btn-orange
}
}
row{
quadruple{
small-title,text,btn-red
}
quadruple{
small-title,text,btn-orange
}
quadruple{
small-title,text,btn-orange
}
quadruple{
small-title,text,btn-orange
}
}
row{
double{
small-title,text,btn-red
}
double{
small-title,text,btn-orange
}
}

@kannangce Do you have encountered a problem seems the result overfiltted? Thank you so much if you can give me some guides.

i have the same problem. did you find a solution ?

[h476564406]

已经收到。

[Yangqi0338]

(py36) E:\koutu\pix2code-xin\pix2code-master\model>python ./build_datasets.py ../datasets/ios/all_data（py36）E：\koutu\pix2code-xin\pix2code-master\model>python ./build_datasets.py ../数据集/iOS/all_data Splitting datasets, training samples: 1499.142857142857, evaluation samples: 249.85714285714286拆分数据集，训练样本：1499.142857142857，评估样本：249.85714285714286 Traceback (most recent call last):回溯（最近一次调用最后一次）： File "./build_datasets.py", line 79, in 文件“./build_datasets.py”，第 79 行，在 assert len(eval_set) == evaluation_samples_number断言 len（eval_set） == evaluation_samples_number AssertionError 断言错误

Why does such a problem occur and needs a solution?为什么会出现这样的问题并需要解决方案？

#40
This solution can fix your problem

[h476564406]

已经收到。