Training in command line

[sjurado]

Hello everyone,
I would like to train 3DMASC using the command line.
I cannot find the command, can someone help me out with it ?

Also in the article, the figure 3 illustrates the classification workflow : there are a 3DMASC Plugin workflow and an additionnal python workflow. Does calling the training by command line enable the python workflow ?
If not, how do you do to use this workflow ?

Thanks for your help

Best regards

[daniel]

Hum, I don't know where the documentation for this lives, but you may be able to extract some info from the code: https://github.com/dgirardeau/q3DMASC/b ... Commands.h

[paul.leroy]

Hi,
Could you give us more details about what you want to do? Did you already have a parameter file and at least a training cloud? Did you already try to use the plugin with the GUI to understand the way it works?

[paul.leroy]

In our Python library, lidar_platform, there are some elements to use 3DMASC with Python. It is a little bit esoteric at the current time but you can find some ideas in the 'classification' directory. Later we will try to add documentation and maybe notebooks for getting to grips with the tool.
https://github.com/Rennes-LiDAR-Platform

[sjurado]

Hi,
Thanks for your replies.

I would like to use the whole workflow (Features computation, Analysis of the correlation between features, Features and scales pre-selection, Classifier training, Classifier optimisation, Application of the optimal classifier to the core points).

From my understanding of the article, I need to use the « Features computation » by command line (« 3DMASC can be called through command line solely for fast feature computation. »). Can you help me to specify the command ? I guess it should be something like : CloudCompare –O « pointcloud.las » -3DMASC_CLASSIFY « classifier.txt » -ONLY_FEATURES
The output of this command can then be used for the « Analysis of the correlation between features ». There again how can I use it ? I’ve looked through the python code but it isn’t obvious to me which function to call.
I have the same question of which function to call for the « Features ans scales pre-selection » step.
Then for the « Classifier training », I suppose I should use the parameters file created from the previous steps and my point cloud and call it using the command line. What would be the command line for this step ?
Then again, how can I run the « Classifier optimisation » ?

To sum up, I would definetly need some hints to use the full classification workflows.

workflow 3DMASC.png (72.38 KiB) Viewed 54159 times

Thanks for your help !

PS : I have used the plugin and I have a cloud and parameter file to train 3DMASC on but I feel like I need to use the whole workflow to properly adress the classification challenge I face.

[paul.leroy]

Did you check the webpage of the plugin? You may find answers to your questions.

https://lidar.univ-rennes.fr/en/3dmasc#p-136