[Qgis-user] counting fish

[Michael Harte]

Jake,

I'm replying in the context of your fish image collection process rather than the image analysis process.

I worked on a project on the Klamath River, in Northern California where we were enumerating salmon stocks with sonar as they ran up the river to spawn.

To ground truth the sonar counts we decided to use video.

We floated a video camera over the ensonified area of the river, but found it difficult to see the fish in the murk of the river, over an algae covered, gravel river bed.

We painted a fiberglass panel white and put it on the bottom to give us a contrasting background, as I assume you are doing.

During post season analysis of the sonar data we noted that the spatial distribution of the fish changed for the two weeks the white panel was on the bottom.

The distribution was uni-modal over the center of the channel for the rest of the season, but bi-modal, with modes at the sides of the channel during the two weeks the white panel was in place.

We concluded that the fish were either avoiding swimming over the white panel or avoiding swimming under the video camera, which was in an large plastic housing floated in a big "inner tube" (we did this in 1987).

However, we did not see the assumed avoidance distribution during the period the camera was in place without the white panel on the bottom.

You might be on the look out for this avoidance behavior.

Cheers,
Mike




> On Feb 6, 2017, at 12:13 PM, Jake Maier <j... at jmforestry.com> wrote:
>
> <image001.gif>
> Is there an easy way, even an existing plugin, to identify photos which
> show an object (fish)  swimming over a white board.
>
> I live on a river and there is a potential problem with alewives, an
> extremely important fish in the food chain. I’d like to find a way with a
> wildlife camera to take a photo at regular intervals over 3 week. The more
> photos I can handle (possibly thousands) the more accurate my count.
>
> I was hoping to find an automated way to identify those photos which show
> an object on the whitish background. I thought it should be possible to
> just look at the average gray value of the photo.
>
> Is there a fairly easy way to do this?
>
> Thanks for any help.
>
> Jake

-- 
Mike Harte