indigowizard / ndvi-viewer Goto Github PK

• Creat NDVI classes
• Classify NDVI raster
• Output new classified NDVI layer

• Identify common colorblindness types
• Add adapted color palette
• Sync color palette between map & legend

The upload_files_proc function is not reading all the geometries within a single GeoJSON file, instead it only parse the first geometry and ignores the rest.

• Fix upload_files_proc function to read through all available geometries within the same GeoJSON
• Read geometries within variant structures of GeoJSON (different from https://geojson.io/ structure)

Error running app. If this keeps happening, please

Needs a section where the user can generate a profile of the data with all the stats related to the results, e.g;

• Size of area of interest in different metrics
• Area surface covered by vegetation
• Area surface of each class of NDVI values
• Percentage of class compared to the area of interest
• Charts for comparison between baseline NDVI and updated NDVI values
• Generating a PDF report with these stats

• optimize css loading for prioritary elements
• added input date in the UI of the layer panel to further differentiate the layers

minimizing input data to avoid cramming UI with x4 date input widgets to form time frames for each satellite image collection.

• Deleting date input label in favor of status text boxes
• Using single date input for each period
• Setting up a 7 days time range calculated from time input and back (time range calculated to relatively to the past of the input date to avoid errors if user picks today's date)
• Creating separate image collections & layers for each of the time inputs

Add legend section explaining the map's elements, includes:

• color palette
• ndvi values
• category info (explaning each color category meaning)

• date input
• convert to date range and hook woth earth engine collection filtering

• Setup requirements.txt
• Setup Streamlit cloud deployement config
• Use Earthengine token as streamlit secret
• Deploy dev branch for testing
• Using geemap.ee_initialize() to initialize/authenticate the gee app in the cloud

Find a way to create comparative maps where each displays the NDVI from a date period specified by user input. Explore;

• Dual map
• [ ] Map slider

The current map is set to arbitrary default coordinates which makes it less user-friendly when uploading aoi and having search the map for it. The ideal way is to implement dynamic map centering mechanic that focuses on the uploaded GeoJSON aoi.

• Calculate center of uploaded geometry list
• Center the map based on the centroid of the geometry
• Use dynamic centering to focus the map on the latest uploaded geometry in case of multiple GeoJSON files

The app is unexpectedly doing good but the more users are trying the app the more it keeps crashing due to Streamlit Cloud free tier limited allocated resources. I cant afford a paid plan on any service so I think I have to focus on app optimization (it needs this regardless of the app crashing or not).

If anyone has any idea on how to optimize a Google Earth Engine app, please check the issue and lend a helping hand.

• Setting up main folium map
• adding secondary vector layer basemaps

• configuring an image collection filtering instead of single ID ee.Image
• setting up collection ID
• clipping collection to aoi

• Adding areal imagery layer basemaps
• Configuring custom basemap: normal & colorblind-friendly
  • Normal Vision: No color vision deficiency.
  • Protanomaly: Red-green color blindness where the red cones in the eye are abnormal.
  • Protanopia: Complete red-green color blindness where the red cones are missing.
  • Deuteranomaly: Mild red-green color blindness where the green cones in the eye are abnormal.
  • Deuteranopia: Complete red-green color blindness where the green cones are missing.
  • Tritanomaly: Blue-yellow color blindness where the blue cones in the eye are abnormal.
  • Tritanopia: Complete blue-yellow color blindness where the blue cones are missing.
  • Achromatopsia: Total color blindness where none of the cones in the eye function, resulting in seeing the world in shades of gray.

• Create a GitHub repository for the project
• Set up and sync the dev environment
• Define project development route
• Define the structure of the project

• Configuring Streamlit app main layout
• Setting up app styles
• Responsive UI for mobile
• Sidebar
  • Title
  • Image/logo
  • Navigation section
  • Social media/contact
• Map legend
• Info section
  • results interpretation
  • scientific explanation
  • data
  • process workflow
• About section
• Contribution section
• Credit section
• Contact section

A feature to either save the resulting raster layers as GeoTIFF files or an option to take a screenshot of the Folium map canvas and save it to local machine would be a great feature.

• Build NDVI function based on image collection
• Render a NDVI raster layer clipped to user aoi based on time input
• Mask the water on the image and showcase only land

• Adding project info including intial submission to GHW Data hackathon, credit and updating the project structure

Allow user to input a value for cloudy pixel percentage which helps have clearer images

• slider input for cloud pixel coverage
• set minimum value >0 to avoid app error

• Setting up Google Earth Engine library
• Configuring Google Earth Engine drawing method
• Adding GEE drawing method to Folium

• Add shapefile upload support to design area of interset. Support GeoJSON shapefile format.

The placeholder of the st.date_input streamlit widget was set to the current day, which turns out to be inconvenient as satellite images aren't always available on a daily basis in the dataset or not captured yet, so when the app starts and tries to retrieve a dataset from the current day that is nonexistent it sends an error and dont display the page:

ee.ee_exception.EEException: Image.normalizedDifference: No band named 'B8'. Available band names: [].

malloc(): unsorted double linked list corrupted

Note: this issue is not the sole reason the app crashes, the heavy load of users also causes the app to crash when exceeding free tier allocated resources.

User input file upload enhancement

• support multiple file upload
• support multiple geometries entry
• encapsulate method in separate function outside of main function for better readability and organization

• NDVI concept
• Sentinel-2 & L2A satellite imagery
• Results interpretation