3
$\begingroup$

This would be used to make little "books" out of edX courses, which I am taking.

I would like to write a small program, which automatically takes a screenshot of a video, when a certain amount of pixel (say 100) "pixel_change" change.

I also have a subRip file, which holds the transcript of the video file. I would like to kind of make a image book, where I have the images of the slide, and on the bottom the transcript for the appropriate slides.

  1. Automatically extract all slides from videos with timestamp.

  2. Read subRip file.

1
00:00:04,359 --> 00:00:07,009
Welcome to this lesson in micro and nano fabrication.

2  
00:00:07,447 --> 00:00:09,873  
The picture behind me shows a colorful SEM image

3  
00:00:10,073 --> 00:00:11,948  
of a bi-morph MEMS activator.

4  
00:00:12,148 --> 00:00:14,872  
And in the next few minutes, I will show how it was fabricated

5  
00:00:15,072 --> 00:00:16,722  
in our clean room at EPFL.

6  
00:00:17,175 --> 00:00:20,188  
Although it does not involve all possible fabrication steps

7  
00:00:20,388 --> 00:00:23,276  
that are nowadays available in advanced MEMS processes.
  1. Assign the slides according to their timestamps to each text block from the subRip file.

  2. Export Textblock and associated slides as one image.

enter image description here

How can I do it?

Thanks for any help!

$\endgroup$
  • $\begingroup$ Is your video streaming or you a recording file? $\endgroup$ – Vitaliy Kaurov Oct 15 '17 at 15:49
  • $\begingroup$ @Vitaliy Kaurov A recorded file $\endgroup$ – james Oct 15 '17 at 15:51
  • 1
    $\begingroup$ My suggestion would be to start with the subRip file, as from there you can easily extract the timestamps (with, e.g., StringCases). Then use that to extract the appropriate screenshots, since you say you already have the video file. $\endgroup$ – b3m2a1 Oct 15 '17 at 18:26
  • $\begingroup$ @b3m2a1 How would you import a subRip file ? I tryed the Import[] function, but it does not work ! $\endgroup$ – james Oct 16 '17 at 7:52
4
$\begingroup$

We can use b3m2a1's suggestion to extract screenshots from the existing video file using the subRip timings for the subtitles. I'll assume each slide is displayed in the video when the subtitle occurs.

The subtitle timings in the subRip file provide a quick shortcut to locate video frames for each slide -- avoiding the need to scan the video for matching frames.

Read a subRip file and locate subtitle timings

SubRip files are text files that contain subtitles and timings for a video. First, read the subRip file into a list, where each item in subRip is a line from the file. I've saved the subRip text from the question in a sample file named "subtitles.srt".

subRip = ReadList["subtitles.srt", String];

Next, extract the lines that have the times each subtitle appears and disappears on the screen. The lines for these times have a specific format that we can use to select them using: Select[subRip,StringMatchQ[#,{__~~Whitespace~~"-->"~~Whitespace~~__}]&]. The subRip times use a comma in place of a decimal point. Fix that with StringReplace[...,","->"."]. Finally, split the times at the "-->" string with StringSplit[..., Whitespace~~"-->"~~Whitespace]. Combining these steps gives the start and end times for each subtitle as strings representing hours, minutes, and seconds.

subtitleTimings = 
 StringSplit[
  StringReplace[
   Select[subRip, StringMatchQ[#, {__~~Whitespace~~"-->"~~Whitespace~~__}] &], 
   "," -> "."], Whitespace ~~ "-->" ~~ Whitespace]

We get these strings:

 {{"00:00:04.359", "00:00:07.009"}, {"00:00:07.447", "00:00:09.873"},  
 {"00:00:10.073", "00:00:11.948"}, {"00:00:12.148", "00:00:14.872"},
 {"00:00:15.072", "00:00:16.722"}, {"00:00:17.175", "00:00:20.188"},
 {"00:00:20.388", "00:00:23.276"}}

Convert the timing strings to seconds

Convert each string with DateObject, and convert the times to seconds. Partition the values in pairs.

timesSeconds = 
 Partition[
  UnitConvert[
   Total[DateValue[
       DateObject[#], {"Hour", "Minute", "Second", "Millisecond"}, 
       Quantity]] & /@ Flatten[subtitleTimings], "Seconds"], {2}]

These are the start end end times for each subtitle.

{{4.359s,7.009s},{7.447s,9.873s},{10.073s,11.948s},
{12.148s,14.872s},{15.072s,16.722s},{17.175s,20.188s},
{20.388s,23.276s}}

Find video frames for each slide and import

The frame rate is available from the video file. We need to know the frame rate to get a frame number from a time in seconds. The frame rate is a value in seconds. Assume a QuickTime video file.

frameRate = Quantity[Import["file.mov", "FrameRate"], 1/"Seconds"];

Using frameRate, compute beginning- and end-frame for each subtitle.

framesStartEnd = Round[timesSeconds*frameRate];

Average the start and end frames. This gives the frames to extract from the video when it's assumed a slide is displayed. Use frameList to get slides from the video.

frameList = 
 Round[(First /@ framesStartEnd + Last /@ framesStartEnd)/2];

slides = Import["file.mov", {"ImageList", frameList}];

Timestamps and subtitles

SubRip allows multiple lines of text for each subtitle. For simplicity, assume one line of text per subtitle.

Get timestamps for each subtitle with:

timeStamps = First /@ timesSeconds
{4.359s,7.447s,10.073s,12.148s,15.072s,17.175s,20.388s}

Get the subtitles for each slide:

transcripts =
 Split[
  Select[subRip, !StringMatchQ[#, {__~~Whitespace~~"-->"~~Whitespace~~__}]&], 
  DigitQ[#] &]
{{1,"Welcome to this lesson in micro and nano fabrication."},
 {2,"The picture behind me shows a colorful SEM image"},
 {3,"of a bi-morph MEMS activator."},
 {4,"And in the next few minutes, I will show how it was fabricated"},
 {5,"in our clean room at EPFL."},
 {6,"Although it does not involve all possible fabrication steps"},
 {7,"that are nowadays available in advanced MEMS processes."}}

The results are slides, timeStamps and transcripts.

$\endgroup$
  • $\begingroup$ Your aswere is perfect !! Really well done! Thank you so much!! $\endgroup$ – james Oct 25 '17 at 13:02
  • $\begingroup$ Very nice. I think I did an answer for the OP where I wrote up a SubRip importer that handles multi-line text, if you want to look that up. $\endgroup$ – b3m2a1 Oct 25 '17 at 15:35
  • $\begingroup$ @b3m2a1 Yes, I saw Import of a SubRip file, and it is a simpler, more robust way to import the subRip file, so now, it's linked. $\endgroup$ – creidhne Oct 26 '17 at 1:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.