UI

[grnolan]

All code is in UI folder. Streamlit is an open-source app framework specifically for machine-learning projects. It opens a browser tab on localhost and updates automatically as code is changed.

To run: streamlit run filename

Streamlit API reference

Features:

• Home page:
  • intro to project, instructions (done)
    • can be found in 'intro.txt'
• Summary:
  • total counts of objects belonging to classes and subclasses (example included, need working model to fully implement)
  • map with interesting objects plotted (implemented, see API reference for clarification)
  • raw data (implemented with sample data, includes class, subclass, timestamp, and coordinates; should add altitude/depth if appropriate)
• Spot Sharks:
  • map (same as above)
  • video frames navigation
    • slider widget is interactive and controls frame selected - may be helpful to change to timestamp slider (haven't found an good way to do that with Streamlit yet)
    • should draw boxes using cv2

Next steps:

• get model running on local machine (finally done :), see Casey's code
• add depth map possibly using ArcGIS Ocean Basemap
• adding counts for subgroups based on model output
• fine tune drawing boxes (code will likely be included with working model)
• add sorting feature to sort objects on map by class