Remote Sensing Image Classification via Improved Cross-Entropy Loss and Transfer Learning Strategy Based on Deep Convolutional Neural Networks.
Ali BahriSina Ghofrani MajelanSina MohammadiMehrdad NooriKarim MohammadiPublished in: IEEE Geosci. Remote. Sens. Lett. (2020)
Keyphrases
- remote sensing
- learning strategies
- cross entropy
- image classification
- convolutional neural networks
- remotely sensed data
- hyperspectral remote sensing
- remote sensing images
- change detection
- multispectral
- remote sensing data
- online learning
- remote sensing imagery
- high resolution
- log likelihood
- hyperspectral
- image analysis
- active learning
- automatic image registration
- satellite data
- image processing
- land cover
- remotely sensed
- hyperspectral images
- satellite images
- remotely sensed images
- maximum likelihood
- remote sensed images
- error function
- image features
- earth science
- evaluation metrics
- feature extraction
- face recognition
- information theoretic
- probabilistic model
- pattern recognition
- earth observation
- high quality
- feature selection
- blended learning