bgTranscribe: Automatically Transcribe Films of Matches (pre announcment)
Posted By: sita In Response To: bgTranscribe: Automatically Transcribe Films of Matches (pre announcment) (Henrik Bukkjaer)
Date: Tuesday, 7 January 2014, at 8:14 a.m.
In Response To: bgTranscribe: Automatically Transcribe Films of Matches (pre announcment) (Henrik Bukkjaer)
Basically you've solved the problems that a lot of people said could not be done!
Really? I cannot believe this. This is not a hard problem in computer vision. E.g. there are currently cars on the market which basically can drive on a highway themselves. There is research for doing so also in towns. There are programs which can recognize certain faces in pictures of masses. This all is way harder than observing a backgammon match. So, the question is not whether it can be done. The question is, whether I can do it, and whether I can do it within max. of 1 person year. I do not want to invest more for the first version.
Basically you have all the building blocks now, and just need to put it together.
I fear it's not that easy. Of course there will be mistakes. The really hard part is to correct mistakes (when you have a move and a roll which do not match, decide which one to believe more in the current situation), and to identify the situation where you cannot correct a mistake, and thus have to ask the user by showing him a part of the film, and asking what is going on.
You could consider going for a known clock or even a tablet/phone app as a clock, to make it easy to identify end of turn. Then put if in the backlog to improve this part, making your solution more robust and flexible in this area at a later point. Simply to get to a usable version quicker.
I'm not sure what you mean here. You are talking about some external device which will give me the exact time when a move ended? I think, this one won't help much, even if it is possible to synchronize this with the film (which would be difficult, too). The time when a user hits the clock does not help me, because he has his hand over the board. I need to watch the hands anyway, to decide when I have a free view of the board, and thus can evaluate the new position. And then, finding the end of the move is not the hardest part. Finding a roll is a bit harder. At least it currently seems so to me, while working on these items. Anyway, both is easy compared to the problems stated above, which still lie before me, and are a good chunk of the estimated 25% failure.
Please don't stop working on your project, because someone else claims to be doing the same! Having (at least) two different approaches and solutions will surely make both better over time!
While this one is undoubtly true, it also kills my business model. ;-) The market is so small, distribution costs would probably be higher than the income, when I have to share it with someone else. I might do it anyway, even if I cannot make money with it. I won't decide this until I know more about the other project.
Messages In This Thread
BGonline.org Forums is maintained by Stick with WebBBS 5.12.