-
Notifications
You must be signed in to change notification settings - Fork 28
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
New feature: image detection strategy "edge" #57
base: master
Are you sure you want to change the base?
Conversation
Hi @simonmeggle, This on the first glance seems an awesome addtition to the library! I need to check this thoroughly which I haven't had time yet. I hope to check this in the coming weeks. |
Hi Tattoo, glad to hear that you find this useful. |
Hi Tattoo, I just want to ask if I can be of any help? Regards, Simon |
It would be awesome to have the edge strategy included in the imagehorzizonlibrary to avoid the problems of compression artifacts while trying to test through RDP environments. |
I I am already looking for such a feature, as I often struggle with (nested) remote desktop connections. |
This function is very good~ I look forward to its appearance |
I also took a look at @simonmeggle new features to the Library. We could really improve our testcases and developing process with the implemented debugger and the edge detection. |
This would be very useful in our environment. Hope it gets implemented soon. |
Happy new year! |
It will be implemented soon, but obviously not in this library (the pr still hangs, no one has ever reviewed it). |
Hi,
with this pull request I want to share the work of my valued colleague @gautamilango and me.
It introduces another strategy
edge
to detect images in situations when too much pixels are different and further loweringconfidence
would lead to false (or no) results.The key here is to pre-process both images (reference image and screenshot) with canny edge detection (https://scikit-image.org/).
This reduces the images to the most relevant parts before they are compared.
It makes image detection extremely rebust against
Debug Image
is a helper keyword and starts a UI (written by Gautam) which gives worthful insights aboutThere is also a detailled debugger. It shows the edge detection results in oder to fine-tune the parameters.
We are already using the edge strategy successfully in production (synthetic e2e monitoring with Checkmk/Robotmk) and it works pretty well.
To make this strategy as integrated as possible, we have put a lot of effort into this, without breaking any existing functionality.
Thanks to ABRAXAS Informatik AG in Switzerland, which spent the ressources to make this possible at all! 👍
That's all for now - we are looking forward to hearing from you.
Any suggestion is highly appreciated! :-)
Best regards,
Simon
(simon.meggle at elabit.de)
PS: See also https://blog.robotmk.org/