Rectlabel-support
Rectlabel-support copied to clipboard
ryouchinsa
We need your feedback to improve RectLabel.
We have an e-mail address to support RectLabel users. [email protected]
Also we have a twitter account to inform new updates. https://twitter.com/rectlabel
We had some feedback on e-mail and twitter. "Awesome tool!! Any chance you could add templates to go into PASCAL VOC format or COCO, this would make it even easier." https://twitter.com/EyeingChicago/status/858799956404183040
If you could find some problems for RectLabel, could you post the problem on this issues page? We can share the problem and how to solve it among our users.
Best regards, Ryo
I have not used RectLabel for a while (since late 2017: see https://github.com/ryouchinsa/Rectlabel-support/issues/18). I'm disappointed now to see that it is unusable to just tag some images without paying. Is there a way to get back the old version ?
We have updated RectLabel since 2017 according to feature requests from our users. RectLabel is worth paying because there are features which reduce labeling time.
I'm disappointed now to see that it is unusable to just tag some images without paying.
Could you restore purchase on the purchase dialog? You can label boxes and save them if you bought in 2017. For extra features, you have to pay because those features were not when you bought.
No I have not bought it, it was early stage and free in 2017. I made some feedback then : https://github.com/ryouchinsa/Rectlabel-support/issues?utf8=✓&q=is%3Aissue+is%3Aclosed+author%3Axvolks+
Could you restore purchase and show us the screenshot of the purchase dialog which appears when you launch RectLabel?
I just started using it today. So far very good. However, I couldn't figure out how to create 'donut'-type of polygons (any shapes with hole somewhere in the middle). Is it possible in RectLabel?
Thanks for reporting the issue. To label donut-shaped pixels, you can use polygon tool with erase mode on the brush dialog.
If you need to export the donut-shaped polygon as coordinates [x1, y1, ...], let us know the desired data format.
We are working on the iPad version now. We have requests for Both Linux and Windows version.
Small feature request. Would be great to be able to select and copy the file name of the image shown. Currently this is not possible, so we end up typewriting it.
@adriantre Thanks for writing the issue. In the next update, you can copy the image file name to pasteboard by clicking on the top-left image file name area. Why do you need to copy the image file name during labeling?
@adriantre The new update version 3.02.1 was released. To show the new update on Mac App Store, press command + R to reload.
- Restored "use difficult tag" feature.
- Improved so that annotations are scaled to the current image size when you save.
- To copy the current image file name, click on the image file name shown on the top-left corner.
Why do you need to copy the image file name during labeling?
Awesome, thanks for notifying me! Sometimes we need to take a closer look at a specific image in some other data flow that we have. A general example could be manual hard example mining.
Hi!
We are using basically only SuperPixeltool for masking images. Only feature I'm missing is the ability to select multiple segments at once. i.e With specific selection tool or just by just going through segments with the cursor. Holding the mouse button or switching this feature on & off with a designated hotkey.
@vhuttunen Thanks for writing the issue.
to select multiple SuperPixels at once holding the mouse button
When we could have some progress, we will let you know.
@vhuttunen
- Improved so that you can select multiple SuperPixels dragging the mouse.
The new update version 3.02.4 was released. To show the new update on Mac App Store, press command + R to reload.
Let us know your feedback.
great software.
It would be extra useful to add:
-option to display numbering of bouding boxes in correspondance to the list
-option to store the confidence at each bounding box of the inference done with Core ML inside the produced XML file, eg. like
- option to navigate using image name instead of image index
- way to select all labels at once (keyboard shortcut), instead of selection them all with the mouse, so that there is a simple way to transfer (copy-paste) all label from one annotated image to the next un-annotated one.
- way to export the image with the coloured bouding boxes as seen on screen inside an image file (jpg, png)
- surely other features I forgot to mention. Best,
Michael
@michaelktran Thanks for your feedback.
option to display numbering of bouding boxes in correspondance to the list
You can use "Show label on the box".
option to store the confidence at each bounding box of the inference done with Core ML inside the produced XML file
If we saved the confidence in the XML file, for what purpose are you going to use?
option to navigate using image name instead of image index
You can use "Search images" by image name.
way to select all labels at once (keyboard shortcut)
You can use "command + a".
way to export the image with the coloured bouding boxes as seen on screen inside an image file
Why do you need the screenshot of bounding boxes?
Best regards, Ryo
Hi Ryo, Would it be possible to have an option that makes annotationns tagged in a picture automatically applied to the next picture when pressing next (or the → key)? This would be extremely useful when annotating frames from videos, since in many videos the same objects can be found in consecutive frames with little or no changes. Best regards, Ire
@ireneu Thanks for writing the issue.
Added "Auto copy from previous image" on Label fast tab on the settings dialog.
We submitted the new update to Apple. When the new update is released, we will let you know. Thank you.
@ireneu
The new update version 3.06.8 was released. To show the new update on Mac App Store, press command + R to reload.
Added "Auto copy from previous image" on Label fast tab on the settings dialog. Fixed the Hotkeys buttons disappearing problem on macOS Monterey. Improved so that when you undo deleting polygons, polygon selections are restored.
Hi, @morestart
We localized RectLabel apps to Chinese, Korean, Japanese, Italian, French, German, Russian, Spanish, Portuguese, Dutch, Polish, Swedish, and Czech languages.
The new update version 69 was released. Please let us know your opinion.
Hi, @morestart
We localized RectLabel apps to Chinese, Korean, Japanese, Italian, French, German, Russian, Spanish, Portuguese, Dutch, Polish, Swedish, and Czech languages.
The new update version 69 was released. Please let us know your opinion.
Great Work!!
When I'm still in the experimental phase and I'm exploring my dataset I only want to annotate certain groups of images. (I'm using fiftyone to explore and filter)
I have a very large dataset of thousands of pictures. It would be very useful to me if I could launch rectlabel with command line arguments to filter by a subset of images. Maybe an array of the image names?
FiftyOne looks a great datasets viewer. https://docs.voxel51.com/
If you could use RectLabel combining with FiftyOne, it would be a great use-case. A search feature using the image file names will be important to RectLabel. Please let us have some time to implement. Including the command line arguments, we will find a convenience way. Another idea is to drag a text file which includes image file names to RectLabel.
We improved so that once opened images and annotations folders when the first launch, from the second launch, you can give command line arguments to RectLabel which image files should be opened. You do not have to wait for loading all images and indexing all annotation files.
open -a RectLabel --args -images 000000000872.jpg,000000180011.jpg. open -a RectLabel\ Pro --args -images 000000000872.jpg,000000180011.jpg
The new version 2023.02.20 was released. Please give us your feedback.
We improved so that once opened images and annotations folders when the first launch, from the second launch, you can give command line arguments to RectLabel which image files should be opened. You do not have to wait for loading all images and indexing all annotation files.
open -a RectLabel --args -images 000000000872.jpg,000000180011.jpg. open -a RectLabel\ Pro --args -images 000000000872.jpg,000000180011.jpg
The new version 2023.02.20 was released. Please give us your feedback.
Wow that was quick!
Works great too. Thank you! This speeds things up dramatically for me playing with large image datasets!
We are working on the iPad version now. We have requests for Both Linux and Windows version.
And are these (linux, Ipad and windows versions) going to actually happen?
