Day #194 - Using Convolutional Neural Networks to Automatically Analyze Aerial and Satellite Imagery

Key Lessons

• Tagging - Segmentation - Object Detection
• Raster Vision is an open source framework for Python developers building computer vision 

• models on satellite - https://docs.rastervision.io/en/0.8/
• CNN and ML Overview
• Data Imbalance issues handled with Data Augmentation (Flip / Rotate)
• Resnets 

• Resnet + Inception + DenseNet
• Upsample and add to K Categories needed to segment
• Sliding Window + Ensembling Models
• NDVI in Satellite Imagery

Happy Mastering DL!!!