Open-source · 2023–
CausalImages
An R package and workflow for causal inference when your key variables live in images — from satellite imagery to biomedical scans to social science measurement.
R package
Remote sensing
Computer vision
Causal workflows
Concept
When images are the data
Many of the most important outcomes are hard (or expensive) to measure directly — but correlate with visual signals. CausalImages is built around a simple idea: make image-based measurement compatible with credible causal identification, transparent assumptions, and reproducible code.
Separate measurement from causality
Validate where it matters
Reproducible pipelines
Workflow
Workflow and results
A representative pipeline: data, models, evaluation, and inference — with the goal of interpretable effect heterogeneity and robust policy insight.
