Looking for someone who can code a (small ?) app for me…

[Toffin Etienne]

Hi there,

I'm a PhD student in Biology, working in Brussels (Belgium). I'm  
looking for a dev who would have some time to spent in coding a small  
video app based on python and gstreamer for my lab. Here's the problem  
I want/have to solve…

I'm working in a lab where the main interest is studying collective  
behaviours, such as collective food retrieval or nest digging in ants,  
group decision-making in cockroaches… that have already been used in  
several topics such as informatics (has anybody heard about ant-based  
algorithms ?). Such studies need two main kinds of data corresponding  
to the macro and micro-level of the phenomenon : collective and  
individual behaviours. Those data allows us to understand how  
complexity emerges from individuals behaviours and interactions. (THAT  
WAS FOR THE SCIENTIFIC BACKGROUND)

(HERE'S THE MORE TECHNICAL BACKGROUND) Basically, we make some of our  
measurements by using some programs to compute nest area (region-
growth segmentation) or to follow big enough insects individuals  
trajectories (tracking) for example.
However, the large majority of individual behaviours have to be  
measured by hand as they need "human assessment" to say wether or not  
the behaviour occurs (a small ant that takes a soil pellet and drop it  
after wandering in the experimental setup). In this case, we don't  
have softwares that help us in getting those informations and we have  
to perform a prehistoric-workflow as following :
- watching the video (on a computer… we can at least do that) and  
waiting for a given behaviour to occur (for example an ant beginning  
to drink at a food source),
- pausing the video when the behaviour occurs and writing the given  
time on a sheet,
- playing again until the next awaited behaviour is seen (the ant  
stops drinking), then pausing and writing again the time on the sheet,
- when video watching is done, entering all the values into a text  
file or something else…
Such way of working is very tiring, boring (and all of those bad  
adjectives in -ing)… If you consider that what I gave to you is a  
simple measurements sequence and that a lot of different behaviours  
and measures can be needed (such as the ant cartesian coordinates when  
the behaviour occurs…) you can easily imagine that it's a hard and  
stupid work…

So I thought about writing a small video program that would be an all-
in-one software :
- video player (with precise controls such as frame by frame playing;  
allowing video or image stacks playing)
- keyboard shortcuts that can be easily defined for a behaviour (for  
example Ctrl-D for "ant begins to drink" and Ctrl-L for "ant stops  
drinking and leaves the food source")
- mouse ability to click on the video where the behaviour occurs (to  
measure the cartesian coordinates)
- recording of the time where shortcuts are entered and the cartesian  
coordinates of the ant showing the behaviour
- timeline where already recorded values are shown as the video file  
is replayed (it's often necessary to make our measurements in several  
times as there can be a lot of ants to observe at once)
- interface allowing video calibration : sometimes we have to film the  
experiment not from above and the experimental setup is then deformed  
on the video (circle becomes an ellipsis). Then we need a simple  
calibration interface to automatically convert the measured  
coordinates in real coordinates
This software would be written with Gstreamer and Python (I've heard  
that it would be the best solution) and could be used on Linux, OSX  
and Windows.

There doesn't exist any FREE software at this time and all the  
existing ones are quite expensive (Noldus products for those who know  
them). And you all know the benefits from free softwares. I know a lot  
of people who work with the same pre-informatic-era methods and would  
appreciate that kind of software…


So, here are my questions :
1. is anybody here interested in writing such program ?
2. does anybody know somedy who can fullfill the question 1. condition ?


Fundamental research is not only made of laboratories that have a lot  
of money to spend in expensive softwares and work with industries… The  
large majority of the labs are quite poor (considering our needs).  
Your help will be very very useful.

Thanks for your replys,

Etienne Toffin
       
-------------------------------------------------------------------
Etienne Toffin, PhD Student
Unit of Social Ecology
Université Libre de Bruxelles, CP 231
Boulevard du Triomphe
B-1050 Brussels
Belgium

Tel: [phone]
Fax: [phone]
Skype: etienne_titou
http://www.ulb.ac.be/sciences/use/toffin.html



-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gstreamer-devel

[Olivier Aubert-2]

Hello

A number of applications, used esp. by researchers in human
interaction,  exist that offer at least the core features that you need
(annotation of video material) : Anvil, ELAN, etc. You may find useful
information in the http://multimodal-annotation.org/ workshop.

I would also like to point out Advene ( http://www.advene.org/ ), the
project that we have been developing for a while, which also fits many
of your requirements and moreover fulfills some more technical points
(free software - GPL, developped in python, using gstreamer). It offers
the possibility to annotate audiovisual documents, either with text,
structured annotation or with SVG graphics, and further process the
annotations (search them, produce documents from them, etc).

Note that the SVG annotation can only be rendered using the VLC player
module for the moment (which does not allow frame-by-frame navigation),
but I am following the discussion about the implementation of
gstgdkpixbufoverlay, which would make it easy to have the functionality
with gstreamer too.

Do not hesitate to contact me if you have any questions.

Olivier


-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gstreamer-devel

[David Schleef]

On Fri, Apr 11, 2008 at 11:22:59AM +0200, Etienne Toffin wrote:

> So I thought about writing a small video program that would be an all-
> in-one software :
> - video player (with precise controls such as frame by frame playing;  
> allowing video or image stacks playing)
> - keyboard shortcuts that can be easily defined for a behaviour (for  
> example Ctrl-D for "ant begins to drink" and Ctrl-L for "ant stops  
> drinking and leaves the food source")
> - mouse ability to click on the video where the behaviour occurs (to  
> measure the cartesian coordinates)
> - recording of the time where shortcuts are entered and the cartesian  
> coordinates of the ant showing the behaviour
> - timeline where already recorded values are shown as the video file  
> is replayed (it's often necessary to make our measurements in several  
> times as there can be a lot of ants to observe at once)
> - interface allowing video calibration : sometimes we have to film the  
> experiment not from above and the experimental setup is then deformed  
> on the video (circle becomes an ellipsis). Then we need a simple  
> calibration interface to automatically convert the measured  
> coordinates in real coordinates
> This software would be written with Gstreamer and Python (I've heard  
> that it would be the best solution) and could be used on Linux, OSX  
> and Windows.

I've needed something similar for testing the encoder in Schroedinger.
Right now, I have a simple python script that plays files, occasionally
pausing and stepping the video forward and backward.  At some point,
I would like to be able to easily annotate individual frames or
segments, including marking areas that are not encoded well.  The
resulting annotations would be fed into the quality control
infrastructure, so automation is a primary goal.

Such a project isn't very high on my priority list, and will only
get bumped up when it starts to block other things that are higher
priority.  I would, however, assist with anyone taking up such a
project.



dave...


-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gstreamer-devel

[黄 勇]

I have  write a similiar application using a mechanism like appsrc, although I implement in videotestsrc. My design is that use mplayer in background  and  use shared memory  to  copy the  mplayer  out frame to  videotestsrc. The reasons are:
1. I find gstreamer seek process is very slow.  Especially when you need do seeking in frame precision.
2. leak of a powerful demux/decoder plugin compared with mplayer. 
3. key control functions has been realized in mplayer, and can be customized.

and I want someone to review my design and give advices.
I think the playback speed is not your application concerns, but maybe it will be a good extension.

> Date: Fri, 11 Apr 2008 10:56:22 -0700
> From: [hidden email]
> To: [hidden email]
> CC: [hidden email]
> Subject: Re: [gst-devel] Looking for someone who can code a (small ?) app for me…
>
> On Fri, Apr 11, 2008 at 11:22:59AM +0200, Etienne Toffin wrote:
> > So I thought about writing a small video program that would be an all-
> > in-one software :
> > - video player (with precise controls such as frame by frame playing;
> > allowing video or image stacks playing)
> > - keyboard shortcuts that can be easily defined for a behaviour (for
> > example Ctrl-D for "ant begins to drink" and Ctrl-L for "ant stops
> > drinking and leaves the food source")
> > - mouse ability to click on the video where the behaviour occurs (to
> > measure the cartesian coordinates)
> > - recording of the time where shortcuts are entered and the cartesian
> > coordinates of the ant showing the behaviour
> > - timeline where already recorded values are shown as the video file
> > is replayed (it's often necessary to make our measurements in several
> > times as there can be a lot of ants to observe at once)
> > - interface allowing video calibration : sometimes we have to film the
> > experiment not from above and the experimental setup is then deformed
> > on the video (circle becomes an ellipsis). Then we need a simple
> > calibration interface to automatically convert the measured
> > coordinates in real coordinates
> > This software would be written with Gstreamer and Python (I've heard
> > that it would be the best solution) and could be used on Linux, OSX
> > and Windows.
>
> I've needed something similar for testing the encoder in Schroedinger.
> Right now, I have a simple python script that plays files, occasionally
> pausing and stepping the video forward and backward. At some point,
> I would like to be able to easily annotate individual frames or
> segments, including marking areas that are not encoded well. The
> resulting annotations would be fed into the quality control
> infrastructure, so automation is a primary goal.
>
> Such a project isn't very high on my priority list, and will only
> get bumped up when it starts to block other things that are higher
> priority. I would, however, assist with anyone taking up such a
> project.
>
>
>
> dave...
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
> Don't miss this year's exciting event. There's still time to save $100.
> Use priority code J8TL2D2.
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
> _______________________________________________
> gstreamer-devel mailing list
> [hidden email]
> https://lists.sourceforge.net/lists/listinfo/gstreamer-devel

---

轻松把Hotmail下载到本地，试试 Windows Live Mail。 立即尝试！
-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gstreamer-devel

[Andoni Morales]

 Hello!


For my degree tesis i´m progranmming an application that can help you. It´s actually oriented to sports to help coachs analysing matchs and rivals. It actually records on a database the time when one of the 20 predifined data ocours and allows the review of one of these data by selecting it.  With a litle bit more work it could be adapted to your needs adding a plugin with your requirements . It´s developed in c# and use gstreamer and of course it´s GPL licensed. Actually the developpment is an a beta state. Next week i´planning to publish the page of the project with the fist beta.
Regards Andoni

2008/4/11, Etienne Toffin <[hidden email]>:

Hi there,

I'm a PhD student in Biology, working in Brussels (Belgium). I'm
looking for a dev who would have some time to spent in coding a small
video app based on python and gstreamer for my lab. Here's the problem
I want/have to solve…

I'm working in a lab where the main interest is studying collective
behaviours, such as collective food retrieval or nest digging in ants,
group decision-making in cockroaches… that have already been used in
several topics such as informatics (has anybody heard about ant-based
algorithms ?). Such studies need two main kinds of data corresponding
to the macro and micro-level of the phenomenon : collective and
individual behaviours. Those data allows us to understand how
complexity emerges from individuals behaviours and interactions. (THAT
WAS FOR THE SCIENTIFIC BACKGROUND)

(HERE'S THE MORE TECHNICAL BACKGROUND) Basically, we make some of our
measurements by using some programs to compute nest area (region-
growth segmentation) or to follow big enough insects individuals
trajectories (tracking) for example.
However, the large majority of individual behaviours have to be
measured by hand as they need "human assessment" to say wether or not
the behaviour occurs (a small ant that takes a soil pellet and drop it
after wandering in the experimental setup). In this case, we don't
have softwares that help us in getting those informations and we have
to perform a prehistoric-workflow as following :
- watching the video (on a computer… we can at least do that) and
waiting for a given behaviour to occur (for example an ant beginning
to drink at a food source),
- pausing the video when the behaviour occurs and writing the given
time on a sheet,
- playing again until the next awaited behaviour is seen (the ant
stops drinking), then pausing and writing again the time on the sheet,
- when video watching is done, entering all the values into a text
file or something else…
Such way of working is very tiring, boring (and all of those bad
adjectives in -ing)… If you consider that what I gave to you is a
simple measurements sequence and that a lot of different behaviours
and measures can be needed (such as the ant cartesian coordinates when
the behaviour occurs…) you can easily imagine that it's a hard and
stupid work…

So I thought about writing a small video program that would be an all-
in-one software :
- video player (with precise controls such as frame by frame playing;
allowing video or image stacks playing)
- keyboard shortcuts that can be easily defined for a behaviour (for
example Ctrl-D for "ant begins to drink" and Ctrl-L for "ant stops
drinking and leaves the food source")
- mouse ability to click on the video where the behaviour occurs (to
measure the cartesian coordinates)
- recording of the time where shortcuts are entered and the cartesian
coordinates of the ant showing the behaviour
- timeline where already recorded values are shown as the video file
is replayed (it's often necessary to make our measurements in several
times as there can be a lot of ants to observe at once)
- interface allowing video calibration : sometimes we have to film the
experiment not from above and the experimental setup is then deformed
on the video (circle becomes an ellipsis). Then we need a simple
calibration interface to automatically convert the measured
coordinates in real coordinates
This software would be written with Gstreamer and Python (I've heard
that it would be the best solution) and could be used on Linux, OSX
and Windows.

There doesn't exist any FREE software at this time and all the
existing ones are quite expensive (Noldus products for those who know
them). And you all know the benefits from free softwares. I know a lot
of people who work with the same pre-informatic-era methods and would
appreciate that kind of software…


So, here are my questions :
1. is anybody here interested in writing such program ?
2. does anybody know somedy who can fullfill the question 1. condition ?


Fundamental research is not only made of laboratories that have a lot
of money to spend in expensive softwares and work with industries… The
large majority of the labs are quite poor (considering our needs).
Your help will be very very useful.

Thanks for your replys,

Etienne Toffin

-------------------------------------------------------------------
Etienne Toffin, PhD Student
Unit of Social Ecology
Université Libre de Bruxelles, CP 231
Boulevard du Triomphe
B-1050 Brussels
Belgium

Tel: [phone]
Fax: [phone]
Skype: etienne_titou
http://www.ulb.ac.be/sciences/use/toffin.html



-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gstreamer-devel

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gstreamer-devel