thank you a lot for your very useful video. I have a question, in the case of seasonality in the area , I have to classify my class in different images , the images in which to classify my classes, need to be merged into one stack? or classify as a stack of different images? thanks
Very helpful. Can you please create a legend for this as well in a separate video or a code. Secondly, can I open the resulting TIF file into ArcGIS for labeling using symbology (Unique values)?
Thanks for this excellent tutorial. I followed it in Google Colab, but I get the error 'No valid training data were found'. Do you perhaps know what the reason might be and what I'd have to fix it?
Hello, Thank you very much for this amazing video, I need a little help, while i was following your script according to my desired values, During Forest test, it is giving error "Forest not defined". I have watched your video several times, but could not find the solution.Could you please give your expert opinion ? thank you once again. ❤❤❤❤❤ you are a Gem
Hi there, great tutorials indeed, wondering if you can tell me if i run classification over one image, is it possible to get classified outcomes from more than one image ?
Possible, if it is from the same study site/ area. If is in other region: Training data points needs to be taken within the study area. If you have sample from one region, choose bigger study area (include region where you have training data and region of interest)
@@AshrafulIslam-cc3lp It depends on the season. For example, in post-monsoon, the cropland could be cultivated and fallow. So, the training data points for cropland will be different in different seasons. Again, for waterbody, the inland water area may change depending on the season. However, the tranining data points for built-up, forests, etc., will be same in all seasons. So the training data points collected in one time can be applied in other time periods, if they indicate same season.
I Really congrat you with this great job,... awesome,... I try to do this with my own self script but de Accuracy overall doesn't work, could you give me a little help with this please?, rhank you on advance.
Thank you so much for this great helpful content. I have a question. Anyone can explain please. Is it not necessary to have the same spatial resolution of all the spectral bands of an image before classification? In above video, the "imgcollection" contains 10 bands of Sentinel-2 including both 10 m and 20 m spatial resolutions. How it affects the output image?
"imgcollection" has been stacked for 4 images for 10m and 20m resolution bands ; although the training points have been extracted for 10m resolution bands (in NIR, Red and Green). So I think it will not affect the output of the image since they are all classified in 10m res. bands
A big thanks for the video. It’s very informative and nicely described. The problem that I am facing while running the script is, FeatureCollection for Training data Band values shows (0 elements, 0 columns). And with this, further, Image (Error) No valid training data were found is on the display. Can you help with this problem?
Be careful defining the feature type, class, etc. [For reference: 17:30 - 17:50]. You may consider sharing your code from 'Get Link' option for support.