Advices in image analysis for a newbie

hello everybody !
I tried to use ImageJ but I haven’t any background in image analysis… :cold_sweat:
So, of course, I haven’t got any satisfying result .
I’ve got this kind of picture (see below). It’s a drone view of a field. We can see crop lines with healthy and ill plants (respectively white ones and black ones).
How can I count easily these two types of plants with ImageJ?
Thanks for your advices!! :kissing_heart:

There are different methds available to identify healthy and ill plants. You could try to threshold the image and then count the thresholded type of plants. You could also apply a kind of template matching:

http://imagejdocu.tudor.lu/doku.php?id=plugin:analysis:feature_finder:start

However for your ease I would tend to statistically anaylze and seperate the plants, e.g. with the Trainable Weka Segmentation:

You will also find several tutorials on YouTube!

Another way to statistically classify the data is to use R with ImageJ within Bio7:

Please also consider to choose the right classifier if necessary for a good description of the features (healthy, ill)

On the Internet also look for tree crown identification for some additional techniques.

1 Like

Thanks for your reply.
I think TWS fit well for my job. I tried to install it but it didn’t work.

I copy the file "Trainable_Segmentation-3.1.1-20160630.154138-15.jar " in the plugin folder and restart ImageJ. When I click on the Weka button, this message appear:

well… I tried in Bio7 with the same process, but the error pop-up is the same too :disappointed_relieved:

I don’t know if the Trainable Segmentation plugin is available as a regular ImageJ1 plugin. @iarganda is the plugin available for ImageJ1, too?

However it is already installed if you download FIJI. So this might be easier:

https://fiji.sc/

1 Like

Indeed, TWS has many dependencies, so I totally recommend using Fiji to avoid the trouble of getting everything installed in ImageJ.

2 Likes

I tried with Fiji, it seems to be ok.
Thanks! :smile:

so now, let’s go with settings…

1 Like