Hand Tracking and Shipping Times

[PG8]

Hello RealSense Community,

Two quick questions:

Is the RealSense D435 capable of Finger Tracking, similar to the Leap?

If it does, does this finger tracking work only within a certain distance? (Given the D435's range is 0.11-10m)

Second, if I order one today, how long will it be until one ships (to Canada)?

[MartyG · Accepted solution]

With the RealSense SDK 2.0 software that the new 400 Series cameras use, many of the functions that were in-built in the old Windows SDKS (hand and face tracking, 3D scanning, etc) are now provided by using SDK 2.0 in combination with other software modules on other software platforms such as OpenCV and ROS.

These platforms usually work with an ordinary webcam rather than a specialized depth camera with inbuilt support for joint / face landmark tracking. So the new D415 and D435 cameras should be able to use tracking with such software without needing in-built support in their firmware for that function.

Regarding availability of the D435: because of very high demand for the camera, it keeps selling out. If you place a pre-order now, you should receive one around March in the next pre-order fulfillment phase, according to the most recent Intel shipping estimate.

In the meantime, you could try using an ordinary RGB webcam with SDK 2.0 and OpenCv to achieve your, since SDK 2.0 supports display of imagery from such cameras. The image below was captured from the in-built webcam in my laptop.

[PG8]

Thank Marty. If I'm correct, it sounds like in order to do finger and depth tracking you can just use any camera with a special SDK.

Would that - or the D435 - do precise finger tracking at its max 10m distance? Sounds extreme for something that would be relatively small at such a distance.

[MartyG]

I would expect the image processing technology of the 400 Series cameras to have an advantage over a common webcam. Any affordable camera though is probably going to have difficulty picking out the fine detail of a hand that is 10 m away.

OpenCV tracking solutions seem to commonly use RGB-D, which is depth information (D) mixed in with the RGB video data. In other words, use of 3D depth cameras such as Kinect and RealSense.

These links look like examples of solutions that uses an ordinary RGB webcam with OpenCV for tracking:

http://pythonopencv.com/simple-handfinger-tracking-gesture-recognition/ Simple Hand/Finger Tracking & Gesture Recognition | World of OpenCV, AI, Computer Vision and Robotics Examples and Tutor…

https://web.stanford.edu/class/cs231a/prev_projects_2016/CS231A_Project_Final.pdf https://web.stanford.edu/class/cs231a/prev_projects_2016/CS231A_Project_Final.pdf

https://medium.com/@muehler.v/simple-hand-gesture-recognition-using-opencv-and-javascript-eb3d6ced28a0 https://medium.com/@muehler.v/simple-hand-gesture-recognition-using-opencv-and-javascript-eb3d6ced28a0

[PG8]

Thank you for your help. I marked your answer as Correct.

On an unrelated note, does the D435 need a USB 3.0 Type-C to Type-C cable, or can it use a Type-C to Type-A?

[MartyG]

Thank you very much.

The USB connectors are only listed as 'Type C' in various documentation sources, so I'd imagine that they're C to C. I don't have the cameras myself to check the cable.