daveluo/zanzibar-aerial-mapping

Open source notebooks to create state-of-the-art detection, segmentation, & classification of buildings on drone/aerial imagery with deep learning # Aerial Mapping with Drones & Deep Learning in Zanzibar, Tanzania ## Motivation: Open source R&D notebooks of all the steps (deep learning and otherwise) to create a state of the art deep learning building detector & classifier from high-resolution aerial/drone imagery. Something like this: ### Interactive version: alpha.anthropo.co ## 7/25/2019 Update: In process of rewriting everything as a series of interactive geospatial deep learning tutorials on Google Colab. **See 1st tutorial published 7/25/2019** for a complete data creation, model creation, inference, and evaluation workflow for building segmentation: - Medium post - Previewable notebook - Open in Colab Prior dev notebooks can be found in /archive with details preserved below the line: ------------------------ ## Results: As of 1/18/2019 (internal val only): | | Mean F1 score | Foundation F1 | Unfinished F1 | Completed F1 | All Buildings F1 | |-----------------------------|---------------|---------------|---------------|--------------|------------------| | Internal Val (grid 042) | 0.728 | 0.718 | 0.755 | 0.710 | 0.796 | Top 2 in WeRobotics' Open AI Tanzania Challenge | | Mean F1 score | Foundation F1 | Unfinished F1 | Completed F1 | All Buildings F1 | |-----------------------------|---------------|---------------|---------------|--------------|------------------| | Final Test (grids 059, 066) | 0.697 | 0.744 | 0.692 | 0.655 | 0.723 | | Internal Val (grid 042) | 0.696 | 0.683 | 0.749 | 0.656 | 0.757 | ## Background: blog.werobotics.org > Maps are absolutely essential for decision support. Knowing where buildings are located is a fundamental input for urban planning, public safety, public health, disaster response, …