question-mark
Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Objectron frame rate and tracking accuracy problem

See original GitHub issue

Hi, I currently working on a project that try to mimic the try-on-shoe app like Wanna Kick using the Objectron.

Now, I already put the shoes into the scene, and using the FrameAnnotation to get Position, Rotation and Scale (I’m not using the scale Vector3 straight, but (x+y+z)/3*Vector3.one to maintain the shoe shape, but I’m not sure if it’s right way to do it)

However, I notice that the framerate on pc and android are both not high enough like Wanna Kick or Lens Studio. (30 PFS or higher) 1640837956001

Furthermore, the tracking is not very accurate, it always has micro jiggling, and some angle of the foot is unrecognizable - either shoe is not showing or the shoe is dancing like crazy

I want to know if the frame rate and accuracy can be improved? and how? or are both inevitable?

thanks.

Issue Analytics

  • State:closed
  • Created 2 years ago
  • Comments:9 (3 by maintainers)

github_iconTop GitHub Comments

1reaction
homulercommented, Jan 4, 2022

I notice that the framerate on pc and android are both not high enough

First, please compare the performance with the official sample. If it is significantly slower than the official one, please tell me where it takes a long time.

I want to know if the frame rate and accuracy can be improved? and how? or are both inevitable?

Please ask it in the official community.

0reactions
homulercommented, Mar 9, 2022

@H3NRILiN

Furthermore, the tracking is not very accurate, it always has micro jiggling

I’m implementing MIN_DETECTION_CONFIDENCE / MIN_TRACKING_CONFIDENCE options (#295) and just tested them with Objectron. At least, it seems that “micro jiggling” you mentioned can be reduced by increasing MIN_DETECTION_CONFIDENCE.

Read more comments on GitHub >

github_iconTop Results From Across the Web

Determining the Necessary Frame Rate of Video Data for ...
The network camera sequence tracking a cyclist has a frame rate of 30 FPS (0%). The frame rate can be reduced by 33%...
Read more >
Objectron: Real-Time 3D Object Detection with Smartphone
Objectron is a new SOTA dataset, recently presented by Google AI, which intended to improve 3D object recognition in videos.
Read more >
Objectron: A Large Scale Dataset of Object-Centric Videos ...
Objectron : A Large Scale Dataset of Object-Centric Videos in the Wild with Pose Annotations ; Epoch, 100, 250 ; Learning rate, 1e-2,...
Read more >
Objectron: A Large Scale Dataset of Object-Centric Videos ...
Each instance's translation and size are stored in metric scale, thanks to accurate on-device AR tracking and provides sparse point clouds in 3D,...
Read more >
Increase Frame Per Second (FPS) Rate in Custom Object ...
Here comes the trade-off between Speed and Accuracy. You can reduce the image size until you receive satisfactory accuracy for your use-case.
Read more >

github_iconTop Related Medium Post

No results found

github_iconTop Related StackOverflow Question

No results found

github_iconTroubleshoot Live Code

Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start Free

github_iconTop Related Reddit Thread

No results found

github_iconTop Related Hackernoon Post

No results found

github_iconTop Related Tweet

No results found

github_iconTop Related Dev.to Post

No results found

github_iconTop Related Hashnode Post

No results found