Skip to content
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

Predictions #7

Open
Nuri-benbarka opened this issue Apr 25, 2020 · 11 comments
Open

Predictions #7

Nuri-benbarka opened this issue Apr 25, 2020 · 11 comments

Comments

@Nuri-benbarka
Copy link
Contributor

Are there any plans to display predictions along with ground truths?

@erksch
Copy link
Owner

erksch commented Apr 25, 2020

Actually I have a secret version on my computer that supports 3D object detection and semantic segmentation predictions but I haven't found the time to polish everything. I can make a branch so you can try it out.

@Nuri-benbarka
Copy link
Contributor Author

That will be great if you can share it and if I found anything that needs polishing I will definitely help.

@jiangzhengkai
Copy link

@erksch Would you like to share your prediction visualization tools?

@erksch
Copy link
Owner

erksch commented Apr 26, 2020

Yes I'll do it asap

@erksch
Copy link
Owner

erksch commented Apr 26, 2020

@Nuri-benbarka @jiangzhengkai

I pushed my code to the branch react and opened this pull request.
I basically recoded everything with React because it is much nicer than painful DOM manipulation.
With the react rebuild also comes support for 3D object detection and semantic segmentation.

There is no updated README yet, but to start the server you just run yarn start. Don't forget to run yarn install after checking out the branch.

Regarding the prediction feature:
As you will see, you can specify a server where the current segment and frame is sent to where you can perform predictions. The server should have a POST /predict route that handles that. The server should then return bounding boxes and the viewer automatically displays them.
I'll make a template ready for a server that does this.

@erksch
Copy link
Owner

erksch commented Apr 26, 2020

Here is a template for a server that would perform 3D Object Detection:
https://gist.github.com/erksch/a21a1197dfe7d9ee74c8e530e48f2b95

@erksch
Copy link
Owner

erksch commented Apr 26, 2020

I really hope that works for you and is not too complicated to get into.
As I said this branch is not finished yet and I am very happy about any feedback.

@jiangzhengkai
Copy link

@erksch Thanks

@Nuri-benbarka
Copy link
Contributor Author

Thanks for sharing your code, @erksch,
I was trying to run the new branch; I ran yarn install successfully, then I ran yarn build and got the following warnings:

Screenshot from 2020-04-28 09-49-36

and then I got the following error in the browser:
Screenshot from 2020-04-28 09-43-52

Are you familiar with this error?

@Nuri-benbarka
Copy link
Contributor Author

Nuri-benbarka commented Apr 28, 2020

Thanks for sharing your code, @erksch,
I was trying to run the new branch; I ran yarn install successfully, then I ran yarn build and got the following warnings:

Screenshot from 2020-04-28 09-49-36

and then I got the following error in the browser:
Screenshot from 2020-04-28 09-43-52

Are you familiar with this error?

I got this error because of the command yarn upgrade which you get as a suggestion when you run yarn start. If you don't make the update, it works fine.

@erksch
Copy link
Owner

erksch commented Apr 28, 2020

@Nuri-benbarka The warnings are just linting, nothing important.
Glad you resolved your error. Although you build the app you can also just use yarn start to get a working version. But maybe the build is more stable for the rendering.

You mentioned that only one frame or segment is loaded, I'll look into this.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants