A Weakly-Supervised Change Detection Technique for SAR Images Based on Deep Learning and Synthetic Training Data Generated by an Ensemble of Self-Organizing Maps.
Victor-Emil NeagoeAdrian-Dumitru CiotecLorenzo BruzzonePublished in: IGARSS (2019)
Keyphrases
- weakly supervised
- change detection
- sar images
- gamma distributions
- training data
- object class
- training set
- remote sensing
- semi supervised
- topic models
- superpixels
- learning algorithm
- relation extraction
- unsupervised learning
- data sets
- decision trees
- training examples
- wavelet domain
- object detection
- named entities
- multiscale
- markov random field
- supervised learning
- information extraction
- object segmentation
- neural network
- human visual system
- active learning
- feature vectors
- high quality