Tuesday, June 17, 2008

Meeting and Distributed Tasks

We started work today after we took a one day rest. We really needed this rest to refresh and get prepared for working in VideoSurgeon. We had a meeting today and distributed tasks on every team member, which will be delivered on Thursday 19th June. These tasks are:
1. Testing the object tracking module: We are using the "Deformable Contour" technique to track the object among frames. It is a well known technique, so we were able to find it implemented in C language. That was followed by writing our code in Java language. Now we need to test it to ensure it provides the required functionality. It will be tested individually first, then it should be integrated into the project framework.
2. Integrating object detection module with the framework and testing it: Object detection module has been individually tested before, so it will be now integrated into the project framework and tested again inside it. We will have of course to solve any problems that might arise after integration.
3. Implementing the frame based summarizing feature: The code for this feature has not been written yet. So we will start implementing it. We already agreed before on the algorithm that will be used. It is based on having a reference frame which will be subtracted from the rest of frames. If the result of subtraction exceeds a certain threshold, the frame will be added to the video summary, else it will not be added to the summary since it does not contain enough changes.
4. Finishing the implementation of the background completion: The code for this module is not yet finished. We need to complete it since it provides an important functionality. We have a Matlab code that implements the technique we are using, so we are converting this code into a Java code in an object oriented style of course. The difficulty we are facing is that Matlab has many ready made functions that provides certain mathematical functionalities, which do not exist in Java. We are trying to overcome this problem by using ready made libraries for Java that deal with matrices and provides mathematical functions to be more efficient than implementing them from scratch.

No comments: