Accuracy Assessment in SNAP

[SAR2016]

Is there any way in SNAP to assess the classification accuracy of two classified map (e.g., confusion matrix in ENVI or PCI)?

Thanks

[marpet]

There is no support for this in SNAP yet.
But you can export the data as CSV for example and do it with external tools. Python, Excel or in ENVI

[khakim]

Dear Marpet,

I also want to assess the lclu classification accuracy from H/A/alpha Wishart classification using SNAP. Could you please let me know step by step if we using CVS file?

Thanks a lot in advance.

[marpet]

I have no knowledge about H/A/alpha Wishart classification. So I’m sorry but I can’t help.

[Smpsilva]

Hi, it seems to me that snap still doesn’t have any way of evaluating the final accuracy of a classification process (MAXVER, Kmeans, RNN,…?) correct? is there any intention to do it soon? thanks a lot.

[ABraun]

It’s not implemented, but you can manually calculate it. An example is given at the end of this tutorial: Landcover classification with Sentinel-1 GRD

[Smpsilva]

Hi ABraun, thanks a lot. I’l try it.
SMps

[ousmane]

I am a new user of the SNAP software, I cannot solve this problem when I do the classification
according to the division of the study area, it takes into account the whole scene.

[marpet]

I’m sorry I don’t understand what your problem is. And I think your question doesn’t fit into this thread. You better create a topic and add a bit of explanation what you have done and what you want to achieve.