New Salig group

[AlexDM0]

Hi Wenjie,

I'm Alex from Almende. I have created a new SaligNL group under which we can maintain the code for SALIG++. I have made you and Feifei owner as well so feel free to use it as the main repo for SALIG. I have forked your project into it as well so you can switch repo's without any problems.

I have also committed the person reidentification code from Feifei on there. I don't have the falling detection code. Could you commit that to the repo in that group?

Could you tell me a little about the integration process of the code you wrote and the one Feifei did? You can contact me at alex@almende.org, so we don't have to continue using these issues.

I have assumed you'd prefer English over Dutch but if this is incorrect please let me know.

Regards,

Alex

[wenjiepei]

Hi Alex,

I just created a repository on GitHub too and just want to invite you to join it.

I wonder have you got my source code for the whole project? If not, you can download the code from GitHub.

I will give a specification about the code very soon.

Wenjie

From: Alex [mailto:notifications@github.com] Sent: 2014年10月14日 15:21 To: wenjiepei/SALIG Subject: [SALIG] New Salig group (#1)

Hi Wenjie,

I'm Alex from Almende. I have created a new SaligNL group under which we can maintain the code for SALIG++. I have made you and Feifei owner as well so feel free to use it as the main repo for SALIG. I have forked your project into it as well so you can switch repo's without any problems.

I have also committed the person reidentification code from Feifei on there. I don't have the falling detection code. Could you commit that to the repo in that group?

Could you tell me a little about the integration process of the code you wrote and the one Feifei did? You can contact me at alex@almende.org, so we don't have to continue using these issues.

I have assumed you'd prefer English over Dutch but if this is incorrect please let me know.

Regards,

Alex

— Reply to this email directly or view it on GitHub https://github.com/wenjiepei/SALIG/issues/1 .Image removed by sender.

[AlexDM0]

Hi Wenjie,

Thanks for inviting me to this group. However, since this is bound to your account, would it not be easier if you start using the SaligNL one? https://github.com/SaligNL/timeseries_classification

I made you owner there too. That way we have an independent repository for SALIG.

Regards,

Alex

[AlexDM0]

Also, it's a bit faster if you only use the code with github. All the Visual studion files are quite big and would take a long(ish) time to sync.

I have updated your forked repo with a new git ignore, that only keeps track of the c, cpp, h, txt and dat files.

Regards,

Alex

[wenjiepei]

Hi Alex,

It is fine.

The name of folder in My dropbox is “SALIG_project_code”, I have shared it with your email mailto:alex@almende.org alex@almende.org, if you didn’t receive it, please tell me. You have to put the folder named “data” under the project to run the program.

Here is a simple specification of the code.

In the main.cpp, you can find that the program is running with two threads in parallel. One is about reading data from Kinect in real time, the other is the classification of activities.

The code for the reading data from Kinect is very simple. The classification of activities is done as follows:

We apply two kinds of features:

1. HON4D feature, which is proposed in the paper: http://www.cs.ucf.edu/~oreifej/HON4D.html.
2. Joint position feature, which is relative position information for 20 joints, which is proposed in the paper: http://research.microsoft.com/en-us/um/people/zliu/papers/joint_modeling_final.pdf.

After extracting HON4D feature and joint feature from the Kinect data, we apply FFT (fourier transformation) on them to obtain the final feature for SVM classifier. Then we train the SVM model with the FFT feature to get the SVM model.

In the test phase, we use the sliding window to test every possible video segment. We first calculate the FFT feature for the video segment in the sliding window, then use SVM to classify it to get the final label.

The program can be run in real time on a normal computer, the classification phase is about 2 seconds delay than the Kinect recording phase.

If you have any questions about the code, please contact me.

Best regards,

Wenjie

From: Alex [mailto:notifications@github.com] Sent: 2014年10月14日 16:07 To: wenjiepei/SALIG Cc: wenjiepei Subject: Re: [SALIG] New Salig group (#1)

Also, it's a bit faster if you only use the code with github. All the Visual studion files are quite big and would take a long(ish) time to sync.

I have updated your forked repo with a new git ignore, that only keeps track of the c, cpp, h, txt and dat files.

Regards,

Alex

— Reply to this email directly or view it on GitHub https://github.com/wenjiepei/SALIG/issues/1#issuecomment-59049773 .Image removed by sender.

[wenjiepei]

Hi Alex,

I forgot to say that I uploaded the training data and svm model data in the dropbox since they exceed the allowed file size (100 MB) in GitHub.

From: Alex [mailto:notifications@github.com] Sent: 2014年10月14日 16:07 To: wenjiepei/SALIG Cc: wenjiepei Subject: Re: [SALIG] New Salig group (#1)

Also, it's a bit faster if you only use the code with github. All the Visual studion files are quite big and would take a long(ish) time to sync.

I have updated your forked repo with a new git ignore, that only keeps track of the c, cpp, h, txt and dat files.

Regards,

Alex

— Reply to this email directly or view it on GitHub https://github.com/wenjiepei/SALIG/issues/1#issuecomment-59049773 .Image removed by sender.

[kosalasubasinghe]

Hi wenjiepei,

Im doing a undergraduate research work on human fall detection using kinect sensor. Id like to get a help from you! can u give me ur email address? here is my one "kosalacool@gmail.com" Thank you!