search

IamMRM / Super-Resolution-Earth-geo-observatory-system-for-sentinel-2-imagery-using-deep-learning-hr-mars-

1 stars 0 forks source link

data processing flow #3

Open hanah-fei opened 3 years ago

hanah-fei commented 3 years ago

can you write a data processing flow about sen12ms

IamMRM commented 3 years ago

can you write a data processing flow about sen12ms

What do you mean by that exactly? According to my understanding, SEN12MS has three types of remote sensing data. The first one is obtained from the Sentinel 1 satellite that is radar data of two spectral bands. The second type is obtained from the Sentinel 2 satellite consisting of 13 spectral bands of different resolutions (10 20 60 meters). And lastly for classification and supervised segmentation there comes landcover data from MODIS (Terra and Aqua satellites).

You can find more information about this data from the research paper on this dataset.

hanah-fei commented 3 years ago

thanks for your answer, in you bag of color cod use this path "SEN12MS/processed/ROIs13_high_roshan_s2_train_patches/ROIs13_high_roshan_s2" how to process sen12ms dataset to your format?

IamMRM commented 3 years ago

thanks for your answer, in you bag of color cod use this path "SEN12MS/processed/ROIs13_high_roshan_s2_train_patches/ROIs13_high_roshan_s2" how to process sen12ms dataset to your format?

Actually, I made a custom training and test set suitable for our experiments. This can easily be done by visualizing the SEN12MS dataset points and comparing them from the Google Earth maps. Such that the shortened dataset contains representations from all over the world and in equal amount. We did this for effective usage of our resources and for much deeper experiments