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.

Feature Tracking Stream Support

See original GitHub issue

Start with the `why`:

Feature tracking when combined with, stereo depth, and IMU support (https://github.com/luxonis/depthai-hardware/issues/8) and/or AprilTags support (https://github.com/luxonis/depthai/issues/133) is quite useful in terms of providing self-localization capability on robotic platforms/etc.

We also have initial prototypes of feature tracking working (like here) so adding this to the DepthAI API is not high technical risk.

Move to the `how`:

Move our prototype Feature Tracking support into DepthAI API and formalize support, likely integrating with IMU results for improved movement/orientation accuracy.

Move to the `what`:

Support feature tracking through DepthAI to enable visual odometry applications.

Issue Analytics

State:
Created 3 years ago
Reactions:5
Comments:30 (3 by maintainers)

Top GitHub Comments

3reactions

Luxonis-Brandoncommented, Jul 16, 2021

Great to know, thanks @kunaltyagi . So one thing we will be experimenting soon is providing access through our API to more CV kernels at a low-ish level. We have no idea if it will work (and if it works, if it will be stable), but the idea would be that if we succeed at this it would be reasonable to port anything that exists in OpenCV to run hardware-accelerated on the SHAVES through these exposed, hardware-accelerated kernels. So we have some pressing things we need to get out first, but we’re expecting in let’s say 2 months or so to start investigating if this is possible. IF it is possible, porting ORB may actually be done at the community level (among other useful ports). CC: @szabi-luxonis

2reactions

nickjrzcommented, Jul 13, 2021

Yes, sorry just saw that comment. That definitely helps, thanks for sharing that. @Luxonis-Brandon

Top Results From Across the Web

Robust Feature Tracking in DVS Event Stream using Bézier ...

In conventional feature tracking algorithms for event streams, the events in fixed time intervals are converted into the event images by stacking the...

Robust Feature Tracking in DVS Event Stream using Bézier ...

In conventional feature tracking al- gorithms for event streams, the events in fixed time inter- vals are converted into the event images by...

Live object tracking - Cloud Video Intelligence API

With Video Intelligence Streaming API you can detect and track objects in live-streamed multimedia data. Real-time analysis of the data is then available ......

Stream layers—ArcGIS Pro | Documentation

Stream layers are feature layers with a stream service as their data source. They reference real-time datasets where the observations are live.

Change Tracking Using Table Streams

A stream object records data manipulation language (DML) changes made to tables, including inserts, updates, and deletes, as well as metadata about each...