3 ms·
Remote | Pair Inc (https://www.pair3d.com https://www.pair3d.com) Senior Computer Vision Engineer Pair Inc. is the leading Augmented Reality commerce platform
by AndrewKemendo 11y ago
Remote | Pair Inc (https://www.pair3d.com https://www.pair3d.com) Senior Computer Vision Engineer
Pair Inc. is the leading Augmented Reality commerce platform on mobile with leading manufacturers and a rapidly growing user base. We are building the infrastructure of the metaverse one customer at a time.
Ideal candidates will have strong familiarity with real-time image processing and embedded vision algorithms. You should have a working knowledge of object tracking, feature detection, multi-sensor fusion, structure from motion, object detection/classification, scene understanding, visual odometry, and 3D scene reconstruction.
Required Qualifications
* MS/PhD in Computer Science, Mathematics, or equivalent
* Strong object-oriented implementation skills (C++)
* Deep experience with OpenCV
* Experience applying machine learning to real-world vision problems
Interview Process (~1 week total):
1. We review any applicable previous work you provide (CV/Portfolio/Github)
2. 30 minute Phone or Skype call with team
3. Paid work sample test specific to position
Send inquiries to: contact@pair3d.com
- zerr 11y agoAny openings for more generalist C++ engineers (willing to learn CV/OpenCV)? Kudos for the interview process!
- AndrewKemendo 11y agoAs long as you can provide immediate value, then yes. Send us a note and we can talk about it.
- 8note 11y agoWhat would be considered deep experience with OpenCV? also, assuming you have said experience, is there a function somewhere that lets you do a findHomography with ransac, but lets you assume there's no perspective transformation? estimateRigidTransform seems close, but doesn't like outliers?
- AndrewKemendo 11y agoYou should be comfortable enough with OpenCV that you can extend or modify it as needed, to accommodate additional use cases that are not within it's current functionality. edit 2: solvePnP should give you what you need for your second question.
- 8note 11y agoI think I've poked around in the source enough to be able to do that -- running into failed assertions in ORB and such:P 2: Thanks! I'll give that a try!