Accessibility Project Update

November 16, 2010

Okay, one month in, and I still feel like I haven’t done enough. I keep hitting the silliest road blocks.

At first, I started working on some front end things and integrating into the chrome browser. Admittedly, I don’t feel like I did enough for a first milestone. I had basic front end functionality and basically no back end functionality.

Milestone 2: ‘Coming Along’

I decided to focus on server side processing and have the client interact with the ‘master server list’ of videos that are problematic. This meant that I could write a server side application that just scrapes YouTube for content and processes each video. Initially, I thought this would be OK. I knew the Python Imaging library existed and so I thought I’d try my hand at it. That was a failing attempt. Not only did I not remember much Python, I couldn’t even get a development environment set up (I’m terrible at that sort of thing). Next, I thought I could do it in Java, since I’m more familiar with Java. Again, I couldn’t get my environment set up. (yes, I’m a failure). The Java media Framework did NOT want to cooperate with me. In the end I turned to Matlab. Trusty, reliable, bloated Matlab. This was because it’s just easier. No environment to set up, it has all the packages I need already so I didn’t have to go out and download/install other libraries. Simple! Also, given that I’d already done this type of thing in Matlab, I had the basic code written up in 15 minutes. *sigh* <– One whole weekend of work….

My Matlab program appears to come up with the same results as the PEAT application, but does not display them as nicely. I look at the average value of each frame and compare it to previous frames. If there is a difference of moret han 100 between a set of frames, then set a warning because this indicates a large change in the sequence of frames. Granted, this is not the most scientific way to detect flashes in a video, but it’s start. To save time, we can stop processing the rest of the video once a warning sign has been detected.

Next I had to set up a local server. I decided to just use my localhost. Googled how to do it, and found IIS simple enough for my purposes. Now my client can connect and read data off my localhost. *good enough*

My new problem is trying to get a modal dialog box to appear in the YouTube window that covers the video and warns the viewer of the video. I can get a regular dialog box, but not the flashy in window frame. I might just stick to this for now.

Next Steps:

Trying to get the server to automatically scrape YouTube for videos to process OR accept requests from the client. Accepting requests however, means that I can somehow automate the process of grabbing videos off YouTube, process it, and return the result back to the user in a timely manner. (not likely to happen anytime soon)

I need to make the client side more usable. Still lots of “features” that need to be attended to. I also need to test on a real server, but I need help setting one up properly and not just hacking together a default localhost.