Segmentation of Small Objects in Satellite Imagery Using Dense U-Net in Massachusetts Buildings Dataset

Keyphrases

• Satellite Imagery 100%
• U-Net 100%
• Small Objects 100%
• Massachusetts Building Dataset 100%
• Urban Areas 50%
• Class Imbalance 50%
• Overall Accuracy 50%
• Semantic Segmentation 50%
• Earth Remote Sensing 50%
• Remote Sensing Image 25%
• Segmentation Method 25%
• Contextual Information 25%
• Downsampling 25%
• Upsampling 25%
• Deep Convolutional Neural Network (deep CNN) 25%
• Object Class 25%
• DenseNet 25%
• Pixel-wise 25%
• Large Object 25%
• Remote Sensing Image Segmentation 25%
• Urban Remote Sensing Image 25%

Earth and Planetary Sciences

• Satellite Imagery 100%
• Remote Sensing 100%
• Massachusetts 100%