Higher zoom and no image smoothing #1128
Replies: 3 comments 2 replies
-
One temporary workaround would be to cut the input images (4000 x 4000) into smaller ones before labelling and processing (many neuronal nets will have memory problems or inbuilt downsampling anyway). With, say 2000x2000, or smaller images this problem would be less severe. |
Beta Was this translation helpful? Give feedback.
-
I've solved both issues. However, my code is pretty dirty: I work with thermographic images and have added some functionality related to these. If you are interested, can give you the code with my corrections. |
Beta Was this translation helpful? Give feedback.
-
Sure (although I’m ashamed withe quality of my code ;)
https://cloud.ee.pw.edu.pl/nextcloud/index.php/s/AMkytNX4XacTaoK
Regards,
Jacek Starzyński
… Wiadomość napisana przez Kentaro Wada ***@***.***> w dniu 25.06.2022, o godz. 11:56:
Can you post a link to your solution? Possibly someone (or me) can fork that code to include in this repo!
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.
|
Beta Was this translation helpful? Give feedback.
-
I'm working on a project with very small features inside of very big images that need labelling. Features may be 3 or 4 pixels wide at some points in an image that is 4000x4000 pixels. It would be nice if I could zoom to say, 2000 percent instead of 1000 percent.
Image smoothing also makes it difficult to correctly place the vertices for polygons correctly. It would be nice to be able to zoom in without smoothing so that I can place the vertices while viewing the actual pixels.
Beta Was this translation helpful? Give feedback.
All reactions