I would like to use Intel RealSense camera to stream and highlight a box's bounds, then to show its size using measurement API.
Example video : https://www.youtube.com/watch?v=mEHaR2mjm9c
I was thinking about using OpenCV first to detect the box and its bounds (I'm not sure if I'm able to do it using RealSense SDK) then using measurement API to display additional information.
But I'm not sure how to proceed the box detection, I'm not an expert into image processing, but I have two ideas :
- Detecting box bounds using quadrilateral detection to analyze box sides.
- Detecting a box corner : 3 lines crossing in the same point, giving 3D plane origin, then using lines' size to imagine box aspect and display it using POSIT algorithm. - 3D pose estimation - Wikipedia, the free encyclopedia
I've got a R200 model camera.
What do you think about it ? Could you provide any useful information ?
RealSense is being used in a Chinese warehouse to measure boxes. I don't have confirmation of what model (F200 or R200) is used, but given the range limitations of the F200 and the need for the camera to look outwards towards the warehouse, the R200 is more likely. An article on the IDF 2015 conference in China in 2015 said:
"[Intel] announced its partnership with JD, a Chinese online retail giant. Intel showed how its RealSense technology can help improve the Chinese company's warehouse management. In a demo using a tablet equipped with RealSense depth camera, the company showed how the technology can easily measure box sizes required for products in various shapes. The demo also showed how it can easily determine the space that is needed for storage or shipment."
Edit: re-reading the quoted text, it says a tablet was used to scan the warehouse. Almost certainly an R200 used, then
Hi Marty G - I've been looking through the documents on the Intel IDF in China and I can't find a reference to the partnership with JD - is it possible that you could post a link?