Output

[dberma15]

I'm not really sure I understand the output. Could you explain the two files that are output, what the columns mean, what the units are, etc.

I received two output files for a 30 second video but the "# time" column in one file goes up to 8 seconds and the "# time" column in the other only goes up to 12 and has 291 rows. Can you explain this?

[dilawar]

Both algorithms computes every time N and M entries have been added to their input timeseries. When the process terminates, last N and M entries are not written to result file. For examples, if your video was 300 seconds long, files will contain (300-24) seconds and (300-18) seconds - given that fps does not change.

I'll leave this issue open to be fixed later.

[dberma15]

I'm sorry but I'm a bit confused by this. What are N and M? Where did you get 24 and 18 from? The webm video I made is a looped video where the person blinks twice two seconds in, then nothing for six seconds, 2 blinks, nothing for six seconds, etc. Why are these the results? Here is a gif of the video I'm using https://gfycat.com/LightheartedSolidAfricanwilddog

eyetracker_test.webm_out.xlsx eyetracker_test.webm_blinks_using_edges.xlsx

[dilawar]

Thanks for the data. I'll test it. N and M were just placeholders, I forgot how long the time series had to be.

⁣Sent from BlueMail ​

On 14 Mar 2017, 6:03 PM, at 6:03 PM, dberma15 notifications@github.com wrote:

I'm sorry but I'm a bit confused by this. What are N and M? Where did you get 24 and 18 from? The webm video I made is a looped video where the person blinks twice two seconds in, then nothing for six seconds, 2 blinks, nothing for six seconds, etc. Why are these the results? Here is a gif of the video I'm using https://gfycat.com/LightheartedSolidAfricanwilddog

eyetracker_test.webm_out.xlsx eyetracker_test.webm_blinks_using_edges.xlsx

-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/dilawar/eye-blink-detector/issues/4#issuecomment-286407327

[dberma15]

What I'd like to get out of this tool is a csv with a list of times into the video indicating when a blink occured. Is this capable of doing that?

I just realized something that might be an insight to the problem. I get the following error:

Traceback (most recent call last):
    File "blinky.py", line 61, in <module>
        main(vars(args))
    File "blinky.py", line 35, in main
        pixalBlinks = extract.find_blinks_using_pixals(data)
    File "C:\Users\bermads1\Documents\SCITE\blink\extract.py", line 140, in find_blinks_using _pixals
        bT, yy = t[x-1:-x], w[x-1:-x] - smoothW
ValueError: operands could not be broadcast together with shapes (0,) (75,)

I'm not sure what I did wrong. the webm file was saved in the same directory as blinky.py as eyetracker_test.webm and the command I run is: python blinky.py -f eyetracker_test.webm

[dilawar]

I tested the your video. Seems like opencv is having trouble converting it to grayscale after first few frames.

 $ ./blinky.py -f ~/Download/issue4.webm

Unfortunately I am short of time these days to test/reply quickly as I wish. Moreover, we are now moving this demo project into a real time system (with lab-rats); so most of the bugs is likely to be fixed in upcoming month.

[dilawar]

No activity.