my_wiki

issues that matter to me. Check the links below:

• Deep Learning and SAR Applications
• Unstructured-change-detection-using-CNN
• satellite-image-deep-learning
• Change-Detection-in-Satellite-Imagery

---

Also check the topic "change detection in satellite images (SAR if possible) using CNN":

my intuition is that pixel by pixel works are not useful anymore, since 1 pixel might have half an object, i.e. information from multiple inputs. We need to understand the image itself, being able to analyze groups of pixels. The best way to analyze groups of pixels is applying filters (i.e. Convolutions) and analyzing the filters outputs, which might contain multiple pixels... Nevertheless it would be cool to be able to analyze pixel wise at the same time that analysing groups of convolutions,... is this possible?

Related with the pixel CNN topic:

• Auto-Regressive Generative Models (PixelRNN, PixelCNN++)
• PixelCNN
• pixel-cnn++
• PixelCNN++
• pixel-cnn