Thesis

In this report, we will be discussing a proposed software solution for a video editing company by the name of Regent’s IT Solutions. Regent’s IT Solutions requested a software-based solution for determining the momentum of a camera from it’s recorded video alone. Regent’s IT Solutions would use this information in conjunction with other custom software to automatically produce videos for the real estate industry.

We discuss the research we have conducted within the field of computer vision. We will also discuss in detail the process we have gone through in developing the neural network based solution. The solution we have produced makes use of optical flow information gathered from video sequences that were provided to us by Regent’s IT Solutions. We used this extracted optical flow information to distinguish a clear difference between when the camera is moving and when the camera is stationary.

We then used this information to train our neural network to be able to accurately detect the difference between moving video sequences and stationary video sequences. We will go into further detail about the development of our neural network and the possible shortcomings our solution may have.