How tristanscloud/sticklemorph works
StickleMorph is a Streamlit tool for predicting and editing stickleback image landmarks. The user selects an image and predictor, calibrates image scale with two points, edits predicted landmarks on a canvas, then filters and downloads coordinates. The graph separates image/model input, prediction and editing, and output. The sampled app shows CSV export; TPS export is documented in the README but its implementation is not visible in the excerpt.
Main entry points: app.py Landmark Editing · predictors Predictor Files
Components 10
Image and Model Input
- Image Files
- Predictor Files
predictors - Streamlit App
app.py
Prediction and Editing
- Landmark Prediction
app.py - Landmark Editing
app.py - Interactive Canvas
app.py
Calibration and Export
- Scale Calibration
app.py - Landmark Filtering
app.py - Coordinate Export
app.py
Other components
- User
Connections 13
- User → Streamlit App: uses
README.md - Streamlit App → Image Files: reads image
app.py - Streamlit App → Predictor Files: loads model
app.py - Streamlit App → Landmark Prediction: runs prediction
app.py - Landmark Prediction → Landmark Editing: provides landmarks
app.py - Landmark Editing → Interactive Canvas: renders landmarks
app.py - Interactive Canvas → Landmark Editing: returns edits
app.py - User → Scale Calibration: sets scale points
README.md - Scale Calibration → Landmark Editing: scales coordinates
app.py - Landmark Editing → Landmark Filtering: provides coordinates
app.py - Landmark Filtering → Coordinate Export: prepares output
app.py - Coordinate Export → User: downloads CSV
app.py - Coordinate Export → User: offers TPS
README.md