The annotation result image didn't show
sean6585 opened this issue · comments
I tried to annotate the image, but the result didn't show in the interface. The box just remained black, so I can't check my pose annotation.
Software edition (6D-PAT): 2.5
Ubuntu version: 20.04
Hey @sean6585,
that's pretty weird. Which dataset is that one again? I recognize it but cannot remember the name right now. Did you check the terminal output to see if there's any hint there what's happening?
Do you have camera matrices in the folder of the images? I guess there is one, otherwise the program wouldn't load the images but maybe the error lies in there since the images are shown in the gallery.
Ah I see the name it's linemod right?
I'm pretty busy right now due to a deadline but I'll try to help you find the error or have a quick look myself some time.
Thank you for replying !
Yes, it is linemod dataset which I downloaded from the official website.
I directly double clicked the AppImage file to open it, so there is no terminal output I can check.
As the picture shown below, I have created a info.json file which contains the camera parameters by following the format.
Hm looking good to me... I'll try to see for myself what is wrong but I'm not sure when I can get to it due to the deadline :/ You could of course try to run the program from source and try to find out what's wrong.
Could you upload your camera info file here? That would make it easier for me.
And which images did you download? Could you also post the link here?
Dataset link:
https://bop.felk.cvut.cz/datasets/
-> LM-O (Linemod-Occluded) -> PBR-BlenderProc4BOP training
(I only used front 3 images of object 01 to test, and and info.json was coverted from info.yml in the object 01 folder)
Camera info file:
info.zip
If I'm not mistaking your camera info file specifies the image file ending as .png but they are jpgs. I manually changed the .png extension in the info file to jpg and was able to see the images. But it might be that I've got different images than you because the images in the downloaded dataset were of the form 000000.jpg, so there were 2 more zeros than in your info file. Are you using a specific part of the downloaded images? I just extracted some images because the dataset is so huge.
Sorry, I might uploaded the wrong file
I used these files down below for testing, please check
01.zip
That's weird, I cannot reproduce your problem. For me the images show up normally. Maybe add {} to your poses.json file, it's not a valid JSON file without at least the curly braces. I added some new code to automatically do that in the program but it's not pushed yet.
Could you also try to run the AppImage from terminal? Using ./[appimage name]. This way maybe the terminal outputs give away what the problem ist.
Errors I got by running in terminal
And this is my OpenGL version
I see, weird that the program works for me like this since it says that varying is deprecated after version 120 and I'm defining version 150 in the shader itself.
I'll try to commit a fix later.
You can try the new release, I modified the shader code and removed the old keywords that were causing the issue.
I've tried version 2.5.1, then another errors show up
That's so weird, my OpenGL never complains. I'll push a new version when I've got time.
thank you for helping.
Ok next try ;) https://github.com/florianblume/6d-pat/releases/tag/v2.5.2
Does it work now finally?
It works finally haha!!!!
Thank you very much!
Nice! :) And thanks for letting me know about the issue, maybe some other people had it as well.
Hi, can i ask you some questions of annotation?
Yes of course, please open a new issue for your question.
Yes of course, please open a new issue for your question.
Thanks for your reply! I miss the problem just like that : when i'm clicking the points on the left image, it's so hard to achieve a success. I must try many times to get a point on the left image, but it's easy to click the points on the right model. Would you miss the problem like that? I wonder if it's only my problem, looking forward to your reply.
Can you please open a new issue for this? When you click on issues, click on "New Issue". I'd like to keep the issues clean and to themselves and will not answer your question here directly. And can you maybe record a little video of what you mean exactly and post it in the new issue (not here!)? So I can see what you mean exactly.
ok, I will open a new issue later.
If I'm not mistaking your camera info file specifies the image file ending as .png but they are jpgs. I manually changed the .png extension in the info file to jpg and was able to see the images. But it might be that I've got different images than you because the images in the downloaded dataset were of the form
000000.jpg, so there were 2 more zeros than in your info file. Are you using a specific part of the downloaded images? I just extracted some images because the dataset is so huge.
it seems currently the app works for png images too.
