Latest News

Dearest Team,

I’ve been invisible for the past few days I know 🙂 but here’re the results I came up with:

1. For the integration task, I almost finished the integration of the application and I found huge problems in all the algorithms ( Thank god for introducing such noisy data that saleh recorded ).

* For the Blink detection algorithm, now it’s working very fine I guess on all of our datasets.
* And the same for Look down detection algorithm.
* For SSVEP, here are the latest ONLINE results:

Online SSVEP Results
====================

With Threshold = 111

1. Window Size = 10
a. 7 Hz ==> 73 % , 82.8 %
b. 7 Hz (2) ==> 77.16 %
c. 7 Hz (3) ==> 59.8 %
d. 7 Hz (4) ==> 75.4 %
e. 10 Hz (2) ==> 59.7 %
f. 10 Hz ==> 69.5 %

2. Window Size = 9
a. 7 Hz ==> 74.6 %, 78.8 %, 81.5 %
b. 10 Hz ==> 64.6 %, 80 %, 66.3 %
c. 15 Hz ==> 49 %, 44 %, 60.7 %, 65.8 %, 51.7 %

After Lifting up the threshold to 200

1. Window size = 9
a. 7 Hz ==> 96 %, 92.156 %,81.4 %, 97 %
b. 10 Hz ==> 87.7 %, 94 %, 89.5 %,96.6 %
c. 15 Hz ==> 71.4 %, 83.3 %,78.5 %, 85 %, 80 %, 86.6 %, 87 %

You may notice that when I raised the threshold, the results became more interesting except for the 15 Hz that is a bit lower than the other 2 frequencies.

Another thing, we need to record our own data as the past recordings of saleh never raised even a small peak anywhere :D.. And also we don’t have datasets for the 19 Hz frequency which represents a direction in our cursor movement application.

About the application, It’s now integrated and can recognize blinks,lookdowns and SSVEP stimuli in parallel.

That’s all I had to say 🙂

Best Regards,
Hani Amr