A multi threaded Python program runs OpenCV SIFT SURF FAST ORB.
This program will get the video from the camera and use SIFT SURF FAST ORB algorithms to draw the keypoint. The result let you compare different algorithms.
It runs under Ubuntu 14.04 and OpenCV 3.1.0.
To run: $ python featuredetect.py
The image shows the result. The SIFT provides the most accurate feature detection, while ORB and FAST requires the least computational time. Consider FAST is not robust to high level of noise, and the orientation is not computed in FAST. I will prefer ORB algorithm to do feature matching in real time problems.