Testing the Model on New Images. How can i add my own images to the trained model and comprare the accuracy.

[HaRiZ0s]

Dear @avishkarsaha ,

First of all, thanks for the excellent work and sharing the code!

It was easy to set up the environment and train the model.

I am interested in testing the model on new images and check/compare the results. How can i add my own images to the trained model? Any kind of information about the dimensions and type of the images used for the model will be useful.

Thank you!

John

[wang-TJ-20]

@HaRiZ0s ，hello, can you share your train weights? I want to train the model, but my machine performance does not support it

[HaRiZ0s]

@HaRiZ0s ，hello, can you share your train weights? I want to train the model, but my machine performance does not support it

output.txt
train_ious.txt
train_loss.txt
val_ious.txt
val_loss.txt

[wang-TJ-20]

@wang-TJ-20 hello, I have seen your val_loss.txt,. I find the val_loss has not decreased. keep it 35 around.

[HaRiZ0s]

@wang-TJ-20 hello, I have seen your val_loss.txt,. I find the val_loss has not decreased. keep it 35 around.

@wang-TJ-20 hello, yes you are right the val_loss hasnt decreased. I havent paid attention to the "experiment results". I am still trying to figure out how i can add my own images and make the visualization code work which @WilliamLucifer proposed in the other issue. I have my email attached in my github profile. If you want you can contact me so that we can help each other.

[wang-TJ-20]

this is my test.py

…

------------------ 原始邮件 ------------------ 发件人: "avishkarsaha/translating-images-into-maps" ***@***.***>; 发送时间: 2022年8月21日(星期天) 凌晨2:34 ***@***.***>; 抄送: " ***@***.******@***.***>; 主题: Re: [avishkarsaha/translating-images-into-maps] Testing the Model on New Images. How can i add my own images to the trained model and comprare the accuracy. (Issue #10) @wang-TJ-20 hello, I have seen your val_loss.txt,. I find the val_loss has not decreased. keep it 35 around. @wang-TJ-20 hello, yes you are right the val_loss hasnt decreased. I havent paid attention to the "experiment results". I am still trying to figure out how i can add my own images and make the visualization code work which @WilliamLucifer proposed in the other issue. I have my email attached in my github profile. If you want you can contact me so that we can help each other. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***>

[HaRiZ0s]

@wang-TJ-20

Can copy your test.py and share it as a .txt ? Maybe .py files in comment section dont work. You can also copy-paste your code such as the did in #4 visualization issue.

[wang-TJ-20]

@HaRiZ0s
test.txt

[danfouer]

@wang-TJ-20 hello, I have seen your val_loss.txt,. I find the val_loss has not decreased. keep it 35 around.

@wang-TJ-20 hello, yes you are right the val_loss hasnt decreased. I havent paid attention to the "experiment results". I am still trying to figure out how i can add my own images and make the visualization code work which @WilliamLucifer proposed in the other issue. I have my email attached in my github profile. If you want you can contact me so that we can help each other.

I have the same Problem , the val_loss has not decreased

==> Validating on 15 minibatches

D:\ProgramData\Anaconda3\envs\TIIM\lib\site-packages\torch\utils\checkpoint.py:31: UserWarning: None of the inputs have requires_grad=True. Gradients will be None
warnings.warn("None of the inputs have requires_grad=True. Gradients will be None")

[danfouer]

@wang-TJ-20 this is my result by your test.py, is that right?

[zacmilan]

@wang-TJ-20 this is my result by your test.py, is that right?

This result looks bad..

[handsomeZhuang]

? Maybe .py files in comment section dont work

I have this problem, and do you know what the workround is?
Thanks!

[handsomeZhuang]

@wang-TJ-20 hello, I have seen your val_loss.txt,. I find the val_loss has not decreased. keep it 35 around.

@wang-TJ-20 hello, yes you are right the val_loss hasnt decreased. I havent paid attention to the "experiment results". I am still trying to figure out how i can add my own images and make the visualization code work which @WilliamLucifer proposed in the other issue. I have my email attached in my github profile. If you want you can contact me so that we can help each other.

I have the same Problem , the val_loss has not decreased

==> Validating on 15 minibatches

D:\ProgramData\Anaconda3\envs\TIIM\lib\site-packages\torch\utils\checkpoint.py:31: UserWarning: None of the inputs have requires_grad=True. Gradients will be None warnings.warn("None of the inputs have requires_grad=True. Gradients will be None")
hi, have you solved this problem?

[handsomeZhuang]

this is my test.py
…
------------------ 原始邮件 ------------------ 发件人: "avishkarsaha/translating-images-into-maps" @.>; 发送时间: 2022年8月21日(星期天) 凌晨2:34 @.>; 抄送: " @.@.>; 主题: Re: [avishkarsaha/translating-images-into-maps] Testing the Model on New Images. How can i add my own images to the trained model and comprare the accuracy. (Issue #10) @wang-TJ-20 hello, I have seen your val_loss.txt,. I find the val_loss has not decreased. keep it 35 around. @wang-TJ-20 hello, yes you are right the val_loss hasnt decreased. I havent paid attention to the "experiment results". I am still trying to figure out how i can add my own images and make the visualization code work which @WilliamLucifer proposed in the other issue. I have my email attached in my github profile. If you want you can contact me so that we can help each other. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>

hi，friend,
how to get the "lmdb" and "split" directory using for trained and test process?
thanks!

[handsomeZhuang]

@wang-TJ-20 hello, I have seen your val_loss.txt,. I find the val_loss has not decreased. keep it 35 around.

@wang-TJ-20 hello, yes you are right the val_loss hasnt decreased. I havent paid attention to the "experiment results". I am still trying to figure out how i can add my own images and make the visualization code work which @WilliamLucifer proposed in the other issue. I have my email attached in my github profile. If you want you can contact me so that we can help each other.

I have the same problem! how to solve it?