Manipulation of the data in hand to get the best results for our needs.
For example, multispectral cameras are really expensive but we can utilize normal RGB vision of stock cameras with standard UAVs to get NDVI calculated. But beware, this is an impossible task. Because on NDVI you have to get RGB and NIR bands to get NDVI results. NDVI literally is NIR-RED/NIR+RED but there is no NIR band on RGB imagery.
“Any sufficiently advanced technology is indistinguishable from magic.”-Arthur C. Clarke
Custom LAKE & ISLAND designs!
Principle Component Analysis is a really important dimension reduction technique to map the vegetation from satellite imagery. Usually RGB & NIR or NRG used to create these but with GAN we can use anything we like.
Transforming RGB to NDVI. NIR or Red Edge.
MIT Dataset for automated semantic segmentation output
Panchromatic images often used to generate pansharpened images to increase RGB bands’ resolution. We can use panchromatic image to directly color it to get rid of the hassle with combining with other bands.
Audi A2D2 dataset, label to image
A2D2 image to label
Various Super Resolution & Coloring Examples
As we said above, transmorphing is now possible.
Also system can produce lake results too! With that we can create #digitaltwin for lakes to predict their future.
İstanbul Ömerli Dam Lake, Durugöl, Büyükçekmece and Ankara Çınarlı Dam Lake GAN
TARSENS 