labelImg
labelImg copied to clipboard
How to check Already labelled images again?
- OS: Ubuntu I have install labelimg using pip. I have labelled all image successfully in yolo format and train my model using those labels. Now I just want to see bounding box of earlier annotated images to see the comparison of ground truth and prediction done by model. Is there any way by which I can load image file as well as it's annotation text file so that I can see bounding box on that image?
Did you solve the question?
I used labelImg for Pascal VOC. On opening the folder which contains the images that have been annotated, the annotations are already present. Did you also experience the same behavior ?
In addition, did you try specifying your save directory ?
@eb642w No. I am not able to get answer to this yet.
@Tanmay-Kulkarni101 I had tried this, but I am not able to get results.
The same problem, after restart I have lost all labels when they actually exist but not displayed.
The same problem, after restart I have lost all labels when they actually exist but not displayed. (Ubuntu)
running with sudo
helps, so it's something about permissions
Hello Team,
Can we do this with LabelImg? I am also looking for a tool to import already labeled images and verify and correct them. Platform: Windows 10
Thanks a lot!
I followed the coment of @vadzimyatskevich running labelImg with sudo
and everything was Ok! I can check all annotations
@veer5551 Did you find a way to open the annotations and verify them in Windows 10. I tried executing labelImg in Anaconda Prompt as admin mode, but that did not work.
You can edit existing annotations by opening the annotations folder through Change Save Dir.
https://github.com/tzutalin/labelImg/issues/639#issuecomment-700672557
I did "chmod 777" on my images and Pascal/VOC xml files, as well as on the folder were they are. And it worked. sudo did not work for me.
You can edit existing annotations by opening the annotations folder through Change Save Dir.
Awesome solution.
remember putting the classes.txt file in the same folder, it solves my problem.