The main goal of the project is to convert a video sequence into a 3-d point cloud that represents the object's essential characteristics and is able to identify and differentiate the object.

1) Students will need to compute the rotational Matrix of the object. They may have to use SVD (singular value decomposition), PCA (Principal Component Analysis), LP (Linear Programming), Statistical Regression, Least Squares, and other techniques.

2) Students will need to then compute the rotation and scale invariant feature vectors (and the range of values for each vector) for the object, and then implement an algorithm using binary search to search through an ordered list to find the intersections of each vector and then further process intersections in priority order, until there is one match or there are no matches. If there are no matches, a new object ID is created and the comparison feature vectors are added to the list. By processing 20-50 test images, the developer will prove that the algorithm works.

3) Students will need to write code that demonstrates the concept in Python and then convert the code into C++ for fast processing on embedded systems.

Students will have daily check ins online, phone and chat communications.