I'm going to work on hand detection using image segmentation. I am thinking of using Kinect, Leap motion or even using normal cameras, but i can't decide which one is better. I wanted to know if anyone has worked with these cameras or participated in any related work and if there is anything i should know.
That depend on your requirements. You can use normal camera with enhancing hand detection algorithms. This method is cheap, but the accuracy and performance are also competitive to other methods. Kinect and Leap Motion are integrated by sensors, so maybe they give better accuracy.
Hi, I have worked with kinect and normal RGB caeras to segment hand (from shoulder to finger tip) for gesture recognition. In my opinion for indoor, short range application Kinect2 is a decent way to segment. You will directly have the skeletons and some hand gesture information also. After getting the depth & the skeleton, you have to do a little bit of work to extract the hand region. But kinect has its limitations. So be careful. If you are interested only to detect palm region, I have heard that leap motion works quite well. But as I have not worked with it yet, may be someone else can give you better idea about it.
That depend on your requirements. You can use normal camera with enhancing hand detection algorithms. This method is cheap, but the accuracy and performance are also competitive to other methods. Kinect and Leap Motion are integrated by sensors, so maybe they give better accuracy.
Both kinect and leap motion are equipped with cameras which are capable of perceiving depth of the scene. When these devices are coupled with their powerful SDKs we can get extremely accurate details of our hands/body parts. Kinect v1 only capable of detecting only the orientation of the hand (full body should be exposed to the device in order to map the skeleton of the hand). Kinect v2 capable of detecting the hand and also register a comprehensive skeleton (even the fingers and all joints) to the hand. Leap motion can only detect the lower parts of the two hands (from elbow to fingers) and register a skeleton like in kinect v2. All these are capable due to the depth information and their SDKs. A generic camera is capable of detecting palm, track the hand, register 2D skeleton and estimate the pose with an accuracy less than 80% in practical situations (numbers are given from my experience, better can be expected in more controllable environments).
You can start with normal cameras, later you can think of other methods.
https://medium.com/@muehler.v/simple-hand-gesture-recognition-using-opencv-and-javascript-eb3d6ced28a0
Check this link
https://www.google.com/amp/s/m.wikihow.com/Read-Palm-Lines%3Famp%3D1
regards
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Human-Computer Interaction