Semantic Labeling of Satellite Images
- Location: SASTRA Deemed to be University
- Project start date: January 2020
- Project end date: May 2020
- Group/Individual: Group
- Individual Contribution: Implemented the early and the late fusion for the given dataset.
- Github: https://github.com/harshithard/Multimodal-Fusion
Description
Deep neural network with SegNet architecture was implemented for multimodal fusion of remotely sensed images. Two kinds of fusion data models, early and late, were implemented. Accuracy of labeling on comparison with a benchmark - 94.5%. The project makes use of image data from satellites of urban regions in Germany.