Day 5

Today began with the usual morning meeting before the interns split into their lab groups. Joe gave Anoop and I a list of tasks to complete so we won't have any gaps in the story we are telling for our final presentation. Overall, we need to address the process and results of the the most recent freshman imaging class, as well as address the requirements for a hockey analytics system. A hockey analytics system must have a way to track the motion of the players in a function of time, a process to identify the team a player is on, and  a way to identify the name of a player. We must address the feasibility of using broadcast cameras as the basis for the system, which means we needed to go through the footage and assess the percentage of the frames that contain the necessary four reference points. We also needed to determine how many frames contain useful information or features that would help an automated system determine the location in the hockey rink. Lastly, we will compare the accuracy of both systems at meeting the requirements. After talking to Joe, I watched the first period of a broadcast of a RI.T. hockey gave versus Bentley. I documented the reference points every five seconds, and discovered that 60% of the frames do not contain enough reference points. This may change however when Anoop and I go to the hockey rink tomorrow to measure some of the markings. After lunch I worked with some of the algorithms that Carl gave us, but still had very far off answers. Anoop and I typed up our abstract, and then I looked through some of the data sets that Joe shared with us. Overall, it looks like the freshman class had some pretty good success with their program, with almost all of the percentages falling over 90%.