Image Segmentation, Training samples collecting, Ground truth and Reference data handling, in a single video tutorial, here ruclips.net/video/EHancD_AWF0/видео.html
Technically it should be a point, but you can convert the raster to points and use them for the confusion matrix, it will have not so much different result, About pure area based confusion matrix, QGIS 2.X has plugin about it. I once made tutorial about it at 2018, just browse my channel to find it
Thanks so much for this video!!! Please I am mapping marginal land in Europe using open source datasets like Sentinel-2 Global land cover(S2glc), Corine land cover, Pan-European High Resolution Layers (HRL) tree cover density, Natura 2000 network , DEM, European Soil Database Derived data etc... please do you have any suggestions on the type of reference dataset i can use for the accuracy assessment or any other suggestions please.
@@GEO2004 Thanks for your reply, I want to carry out the accuracy assessment for the marginal lands after using the mentioned datasets to mapping them.
I dont know if marginal land is theoretically logic to be evaluated using accuracy assessment, my opinion is it isnt. It is because marginal land is a concept, it is not a physical features in earth surface like landcover or terrain. Marginal land is a complex concept that can be characterized by few or many physical features like soil condition, terrain slope, vegetation cover, climatic condition etc. By this way, marginal land cant be measured directly in the field like true physical features, so in my opinion this concept cant be evaluated using accuracy measurement. And because it is a concept, it lies in the conceptual/perspective level, it is very subjective, you and me probably have different opinion about what marginal land is. But on the technical side, if you still think that marginal land can be evaluated using accuracy assessment, what you can do is try to search official marginal land map published by official/government agency in your area of interest, and use it as reference data for the accuracy assessment
@@GEO2004 Thank you so much for this detailed reply but please do you have any idea to give me on concepts i can use for performance evaluation after the whole theoretical mapping process.
I have a reference map, and i want to match its class name and class value with classified one. but reference map in uneditable. can you help me to solve it, please?
@@GEO2004 thank you for your answer. it is exported from Envi in "Erdas Imagin" format and its extension is ".img". should it be changed? I am sorry, I am new to this.
@@GEO2004 I don't know how to thank you for helping me. can you recommend me any training video on this matter on youtube? I tried with one but the result was odd!
Image Segmentation, Training samples collecting, Ground truth and Reference data handling, in a single video tutorial, here ruclips.net/video/EHancD_AWF0/видео.html
How to get reference classified data of years??
Field survey, higher quality data, government data, etc
How we can make references data ?
I think it is already clear from the video, you can use georeferenced raster image
Yeah, the video just an example, you can always doing the analysis using your own data for best result
how to generate confusion matrix in arc map
It can be done, but the processing steps would be very long, I will try to make one in the future
How about area-based accuracy assessment? can ArcGIS Pro do that?
Technically it should be a point, but you can convert the raster to points and use them for the confusion matrix, it will have not so much different result,
About pure area based confusion matrix, QGIS 2.X has plugin about it. I once made tutorial about it at 2018, just browse my channel to find it
Thanks so much for this video!!!
Please I am mapping marginal land in Europe using open source datasets like Sentinel-2 Global land cover(S2glc), Corine land cover, Pan-European High Resolution Layers (HRL) tree cover density, Natura 2000 network , DEM, European Soil Database Derived data etc... please do you have any suggestions on the type of reference dataset i can use for the accuracy assessment or any other suggestions please.
Can you be more specific about what kind of accuracy assessment you are going to do? It is for the marginal land or for the mentioned landcover data?
@@GEO2004 Thanks for your reply, for the marginal lands after using those datasets to map it.
@@GEO2004 Thanks for your reply, I want to carry out the accuracy assessment for the marginal lands after using the mentioned datasets to mapping them.
I dont know if marginal land is theoretically logic to be evaluated using accuracy assessment, my opinion is it isnt.
It is because marginal land is a concept, it is not a physical features in earth surface like landcover or terrain. Marginal land is a complex concept that can be characterized by few or many physical features like soil condition, terrain slope, vegetation cover, climatic condition etc. By this way, marginal land cant be measured directly in the field like true physical features, so in my opinion this concept cant be evaluated using accuracy measurement.
And because it is a concept, it lies in the conceptual/perspective level, it is very subjective, you and me probably have different opinion about what marginal land is.
But on the technical side, if you still think that marginal land can be evaluated using accuracy assessment, what you can do is try to search official marginal land map published by official/government agency in your area of interest, and use it as reference data for the accuracy assessment
@@GEO2004 Thank you so much for this detailed reply but please do you have any idea to give me on concepts i can use for performance evaluation after the whole theoretical mapping process.
I have a reference map, and i want to match its class name and class value with classified one. but reference map in uneditable. can you help me to solve it, please?
Did your reference map come in raster format? If it is yes, then you must convert it to vector data first
@@GEO2004 thank you for your answer. it is exported from Envi in "Erdas Imagin" format and its extension is ".img". should it be changed? I am sorry, I am new to this.
Yes, it is a raster, unless you convert it to vector, you can't edit it
@@GEO2004 I don't know how to thank you for helping me. can you recommend me any training video on this matter on youtube? I tried with one but the result was odd!
Just search for "raster to vector conversion" on youtube, you will find a bunch of videos about that matter.
how we can get reference?
I have an example how to get reference data, check this video ruclips.net/video/EHancD_AWF0/видео.html
Can you please show how to take the refence data
I think I already answered it in another replies
Thanks. But if you made a separate video on then it way help me alot.
@@jannatunnimenishat6835 okay I will do that when I have spare time,
already done in this video ruclips.net/video/EHancD_AWF0/видео.html