eCognition Deconstructed: Generate labeled sample patches
Learn about how to use the "Generate labeled sample patches" algorithm especially looking at the implemented Sample Augmentation parameters since v10.3 in the Trimble eCognition Software.
This algorithm creates sample image patches and exports training samples with a label that can be used for training of a convolutional neural network in eCognition. The image patches for the sample generation are randomly selected from all pixels in the image object domain and exported and saved to a selected folder. Optionally, sample augmentation can be applied to expand the training samples artificially to improve performance and ability of the model.
Please find more information, data & rule sets and discussions in our eCognition Knowledge Base: https://support.ecognition.com/hc/en-us
_____________Video Content________________
00:00 - Introduction & Theory
01:00 - Hands-On
04:40 - Sample Augmentation