I would like to ask - why I receive slightly different results for OBIA by shapes and OBIA by grids? - I defined five training signatures - in "shape version of OBIA" I received an assignment of all segments to my five categories - in "grid version of OBIA" I received six categories - one category contained segments that were not assigned to my signatures - why - is it not same algorithm, method?
I can try to answer. In grid you may have adjacent pixel value while in shape one tries to create precisely boundary. Statistically this would change the result a bit. hope it satisfies.
Hello. Is there some way that I can update the 'Training' attribute field for all selected polygons after selecting several of them? It's a bit time consuming only being able to update one at a time. It's quite easy in other GIS software but I can't figure it out in SAGA. Thanks for any help
Thanks for answering my question. I'm trying to use your methodology to classify vegetation types using very high resolution orthos from a UAV so updating the attribute field for each polygon is very time consuming. Something else I was wondering is if there is any documention on the 'Supervised Classification for Shapes' tool? I would like to understand how it works of possible. Also do you have any means of generating a classification accuracy for your methodology? I have just seen your new tutorials website that you launced 4 days ago. These look really great!
I have seen that these algorithms are within SAGA so it might be possible. The alternative might be to carry out this part of the processing using R or Python but that is beyond me at the moment.
how do you differentiate between mangrove and any other vegetation so the computer won't confused which mangrove and which is another vegetation? im following your tutorial step by step by the way
Choose a band combination for the classification that works best for picking the mangrove. You may set some classification error that you can edit manually.
![[TH] RNT vs SPG - VCT Ascension Pacific - Day 1](http://i.ytimg.com/vi/RUTkf8Detj8/mqdefault.jpg)
Hello, thanks for this great tutorial. In the end, how to calculate the land cover area in meter square or km2 ?
I think this is explained in the most recent video showing OBIA with Sentinel data.
I would like to ask - why I receive slightly different results for OBIA by shapes and OBIA by grids? - I defined five training signatures - in "shape version of OBIA" I received an assignment of all segments to my five categories - in "grid version of OBIA" I received six categories - one category contained segments that were not assigned to my signatures - why - is it not same algorithm, method?
I can try to answer. In grid you may have adjacent pixel value while in shape one tries to create precisely boundary. Statistically this would change the result a bit. hope it satisfies.
how to determine the accuracy of Classification Test Results ???
Hello. Is there some way that I can update the 'Training' attribute field for all selected polygons after selecting several of them? It's a bit time consuming only being able to update one at a time. It's quite easy in other GIS software but I can't figure it out in SAGA. Thanks for any help
This is a good question - I have not yet found a way to do this. This is worth posting to the SAGA source forge forum.
Thanks for answering my question. I'm trying to use your methodology to classify vegetation types using very high resolution orthos from a UAV so updating the attribute field for each polygon is very time consuming.
Something else I was wondering is if there is any documention on the 'Supervised Classification for Shapes' tool? I would like to understand how it works of possible.
Also do you have any means of generating a classification accuracy for your methodology?
I have just seen your new tutorials website that you launced 4 days ago. These look really great!
Thanks for this. Is there any way to do exactly the same segmentation but the classification with random forest or SVM?
Possibly - I have not tried this.
I have seen that these algorithms are within SAGA so it might be possible. The alternative might be to carry out this part of the processing using R or Python but that is beyond me at the moment.
I recommend a post on the Saga gis source forge forum. They would answer this question directly.
how do you differentiate between mangrove and any other vegetation so the computer won't confused which mangrove and which is another vegetation? im following your tutorial step by step by the way
Choose a band combination for the classification that works best for picking the mangrove. You may set some classification error that you can edit manually.