Deep Learning Workshop for Satellite Imagery - Training & Prediction

Dive into the second part of a comprehensive three-part workshop on deep learning for satellite imagery. Learn to train deep learning models, debug locally, and make predictions with performance evaluations. Explore the U-Net model architecture, implement custom loss functions, and master model metrics. Practice code debugging, model training, and prediction techniques. Discover how to save and export models, create network plots, and use diagnostic callbacks. Visualize your model with Netron and access the source code on GitHub for further study. Perfect for those looking to enhance their skills in applying deep learning to satellite image analysis.

- Content Starts
- What is covered?
- Part 1 Refresher
- Part 2 Coding Starts
- Jaccard Index
- U-Net Model Architecture
- DL Network Coding Starts
- Model Metrics
- Code Debugging 1
- Custom Loss function
- Model Training
- Code Debugging 2
- Model Metrics/Eval
- Model Prediction
- Model Prediction
- Save/Export Model
- Model Network Plot
- Diagnostics with Callback
- Model Visualization with Netron
- Source Code at GitHub