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

Can't make a prediction #1012

Closed
IhorKosovych opened this issue Mar 31, 2019 · 2 comments
Closed

Can't make a prediction #1012

IhorKosovych opened this issue Mar 31, 2019 · 2 comments

Comments

@IhorKosovych
Copy link

Thanks for your awesome code share and explanation!

I trained the model. The performance was good.


step 10 - loss 119.7686767578125 - moving ave loss 119.82690792130654
Finish 10 epoch(es)
Checkpoint at step 10
Parsing cfg/yolov2.cfg
Loading None ...
Finished in 5.650520324707031e-05s

And then I tried to predict another image.
But after

print(tfnet.return_predict(original_img))

The result is:

[]

But instead of empty brackets should be image prediction.

Can somebody help me?

@IhorKosovych
Copy link
Author

The solution is to increase number of epochs. In previous case I set epoch - 100, in next case - I changed epochs to 2500 to get more then 10000 steps in result. If you get more then 10000 steps you will receive image prediction.

@ankitAMD
Copy link

ankitAMD commented Jan 3, 2020

The solution is to increase number of epochs. In previous case I set epoch - 100, in next case - I changed epochs to 2500 to get more then 10000 steps in result. If you get more then 10000 steps you will receive image prediction.

@IhorKosovych
i am training on 1 custom image detection with epoch 1500 loss becomes 1.05 something .
when i put threshold value 0.3 or 0.5 or 0.7 there is no detection .
so i change threshold value 0.03 ,0.01 on different image then it detect but properly detect.
so what i can do for better result.....please guide me.

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