An Industrial Think Tank Focused on Developing and Promoting AI Technology for Geospatial Applications [Please note this group was formed as an academic exercise for educational purposes and does not represent a real world organization]
MIT License
7
stars
3
forks
source link
Move Functions found in Project_Zostera.ipynb into code base #148
Our new experimentnearly completes all the requirements for project Zostera.
To achieve our milestone incorporating tiff and shape files into our classification system - we should include the following functions found in the notebook:
clip_raster()
snap_raster()
add_features_from_vec_col()
Notice in the notebook that the classifier is done directly with gradient boost (not our classification function). This is because the shapefile represents a small portion of the image data points. How our classifier tool core.simpleClassifier() currently in the code is trained - it expects all cells of an image to have valid classifications (ie a Google Satellite tile <-> a google map tile). Furthermore, our core.classifyImage(), which applies our classification, currently expects class code arrays which represent colors. This led to some rather gross code in the notebook to We should also consider implementing:
core.simpleClassifier(where=some_condition) - where we may omit certain class codes while training the classifier
core.setColors() - where an array of class numbers can be mapped into an array of nice colors.
Your mission: discuss and determine a path forward to implement the above functionality cleanly and concisely into the groups codebase. We would like to enable the general functionality to generate a classifier from a given external image and shapefile.
Our new experiment nearly completes all the requirements for project Zostera.
To achieve our milestone incorporating tiff and shape files into our classification system - we should include the following functions found in the notebook:
Notice in the notebook that the classifier is done directly with gradient boost (not our classification function). This is because the shapefile represents a small portion of the image data points. How our classifier tool core.simpleClassifier() currently in the code is trained - it expects all cells of an image to have valid classifications (ie a Google Satellite tile <-> a google map tile). Furthermore, our core.classifyImage(), which applies our classification, currently expects class code arrays which represent colors. This led to some rather gross code in the notebook to We should also consider implementing:
Your mission: discuss and determine a path forward to implement the above functionality cleanly and concisely into the groups codebase. We would like to enable the general functionality to generate a classifier from a given external image and shapefile.