MAKHFI.COM
Fascinating World of Neural Nets  
Neural Network Forums
Home PageHome Page : General Neural Networks Topics : ANN to solve weighing people as they move...
  You are currently not logged in. You can view the forums, but cannot post messages. | HOME | Log In | Register | Search | Help
Post a Reply on This Topic Post a Reply on This Topic

Author Topic: ANN to solve weighing people as they move...
Russ_T Posted: 22-Oct-10 02:23
  Edit Edit
 
Email the Author Mail   View Author's Profile Profile  
Hi all, I've got a little project I'd like to solve with an ANN if I can and it's the right approach.

Here is the problem I have

We weigh people as they walk across a platform, this only takes a few seconds. We have an analog output that is converted to digital of the weight reading over the short time period. As you can imagine this varies as the person bounces along, but we should be able to get a reasonable average and get an accuracte enough weight.

I need to develop an algorithm to do this.

5 years ago I was at uni using Matlab for my disseration to work on adaptive classifiers and that was ok but I've forgotten a lot in the mean time.

I was wondering, if I created a dataset of correct weights and the output from the scales (so effectively the data and expected result) could I create a ANN instead of my averaging algorithm?

I don't know how accurate this approach would be, and ultimately the software needs to be written in C and put on a small chip. But I like the idea of it so would like to investigate futher.

If anyone has any ideas on this approach, or maybe another forum I can discuss it on, I would appreciate it. I'd like to get back into this kinda thing and it looks to be a fun project.

Thanks
 
Nikola Posted: 31-Jan-11 03:43
Delete Delete    Edit Edit
 
Email the Author Mail   View Author's Profile Profile  
Here's a few things that might affect your estimates:

1. Clothes. Mine can sometimes add 2 kg on my weight.
2. Shoes. These can affect the movement pattern of a person and you might get different input data if someone is walking barefoot or in high heels.
3. Carrying something. If a person is carrying bags (such as groceries), or people (perhaps their children), this will naturally affect their weight, but also their movement patterns.
4. Movement. You haven't specified how this would be used, so I'm assuming a very general case. In case someone just happens to stop and wait on your platform, this might completely mess up your data. Similarly, if someone hops while moving.
5. Disabilities. People with disabilities may have drastically different movement patterns than what you might expect from healthy people.
6. Multiple subjects. If the platform is large enough to hold two people, then it might happen that two people will walk across at the same time, or one will leave the platform and the other enter in such a short period that you won't be able to tell one person from another without additional sensors.

In case of one person with no additional load simply walking across the platform, my first thought is to feed the network with only the minimum, maximum and average measured weight and see how that works out. Somehow, though, it seems to me that simply averaging out the weight curve that you get (without the usage of neural networks) should be correct enough, but that depends on what you need the information for.

Cheers,
Nikola
 
aditi Posted: 03-Feb-11 03:42
Delete Delete    Edit Edit
 
Email the Author Mail   View Author's Profile Profile  
hey frndz.....is dere anyone who can help me in learning the neural network.plzzz
 

Post a Reply on This Topic Post a Reply on This Topic
 

Copyright © 2001-2003 Pejman Makhfi. All rights Reserved.