All entries for November 2006
November 26, 2006
UPDATE : Final version online at the address below (front page prettified thanx to Hoi :-) )
have tidied up but this is till too long, thought i’d put it up for feedback anyway
Report upload is imminent!
its taking a wee while to put together – its currently 400 words too long and I’m just trying to cut it down to 1,500
Maybe be done in about an hour…
Well here’s the final post for my stuff: -
Also a reminder to everyone that you need to do a 250 word piece of “individual reflective writing”. Can everyone post this online by tonight so we can put it on the cd.
Hi – like i said in a comment below, I’ve started collating all the work thanx for putting online. Will write the missing bits too.
Will post +/ email it out for everyone to look at when done.
I’m going climbing for a few hours but heres the Implementation, most of the content is there. Just maybe some more headings and rewording is required. Please take a read and add your own touch.
I will be back later to sort out the code, could someone please start to collate it all together and add the final touches.
I will send out some txts,
November 25, 2006
I have been looking at the implementation and starting to get some ideas on what to write.
I will have these completed for early tomorrow and then get the code together in the evening.
Here is a flow chart that I have made to go with the implementation and is the idealised version that if we had time the program would follow.
Here’s a more concise version of the colour, histogram and blob and contour based tracking parts that Nick wrote. You may need the Microsoft Equation 3.0 component of office installed if it isn’t already, it’s part of the office tools.
Also an update on the bibliography with 2 more sources and numbered, incase anyone wants to reference them in the report.
The word count for the colour + histogram + contour is around 530 words, and the word limit is 1500 so there’s a fair bit to fill up. I may be changing these 3 depending on whether we need to fill up the word count, so they’re here just to give you an idea of what I’ve been doing.Things that need to be finished are: -
- Introduction to the problem
Combining the 3 methods Our implementation (Ant’s doing this)
- Appendix (code needs more commenting)
- Putting together all the parts and printing (do we need colour prints?)
- Creating the CD with all the code, pictures, videos and presentation (should we include the robot resource cd?)
- Signing the contribution sheet
Deadline is Monday, so all of this needs to be done tomorrow!
Here is my contribution. Keep everyone posted on progress!
Description of Further possible Developments to the Project:
Four main ideas were identified to expand and/or enhance the project. The following ideas were decided upon based on what would improve the use of the system for the user, by making it easier to use or improving its accuracy or functionality.
The first suggestion is to improve the speed of the algorithm used, so that the system will be capable of tracking a person who is travelling at a much higher speed than currently catered for. This will involve speeding up the running time of the algorithm in question, possibly by removing more of the redundant data being processed, or by finding a better method of comparing models.
The second suggestion is to introduce functionality capable of tracking multiple people within the same image. This would mean that the two people would have to wear different coloured clothing, or some method should be implemented to distinguish between them.
The third suggestion for improvement would be to explore the problems in tracking a person wearing a multi-coloured shirt further. The aim of this would be to tweak the histogram algorithm already in place to remove redundant background information picked up as part of the blob, thus making the method more accurate.
The fourth suggestion is to effectively remove the initial start-up phase which is currently required to create the blob mask. To achieve this, the system would need to be capable of dynamic person recognition, whereby it could recognise and create a mask of a person anywhere within an image, and continue to track that person based on the model created from this mask.
November 24, 2006
If you’d like me to do any more of the essay please ask… else i’ll just help out with the editing on sunday evening with you guys. Can whoever else who is doing the essay please let everyone know what parts they will be doing…
November 23, 2006
...tis for the group essay.
I’ll start the essay off if that’s ok, then if a couple of other people can expand on it after that i think that’s probably the fairest approach.