Enhancing the KidSat Model: Integrating Geographical Encoding and Data Quality Assessment for Childhood Poverty Prediction
For researchers and practitioners in satellite-based socioeconomic prediction, this work provides a scalable and principled recipe to improve accuracy using only publicly accessible data, though the improvements are incremental over an existing framework.
The paper enhances the KidSat model for childhood poverty prediction from satellite imagery by improving data preprocessing, image quality assessment, and geographic encoding, reducing MAE from 0.2167 to 0.1759 (18.83% relative reduction) on cluster-level severe-deprivation proportion, and achieving an overall MAE of 0.1658 when extended to 33 African countries.
Accurate poverty mapping using satellite imagery is often hindered by (i) noisy and sparse survey-derived supervision, (ii) image quality issues such as cloud cover and image corruption, and (iii) lack of explicit spatial structure in image-only models. Building on the KidSat framework, we develop an enhanced pipeline that improves predictive accuracy via refined data preprocessing, systematic image quality assessment, and mathematically defined geographic encoding. First, we refine the fine-tuning target matrix by resolving high-cardinality sparsity and reducing one-hot dimensionality from 103 to 51 via DHS re-aggregation. Second, we introduce a simple two-stage quality-screening procedure to filter heavily clouded or corrupted observations. Third, we fuse DINOv2 visual embeddings with Spherical Harmonics (SH) location features. Across extensive experiments, these changes reduce MAE from 0.2167 to 0.1759, corresponding to an 18.83% relative reduction on the cluster-level severe-deprivation proportion scale. When extended from 16 to 33 African countries, the best-performing configuration achieves an overall MAE of 0.1658. We find that SH features consistently improve performance over the image-only backbone, whereas higher-capacity coordinate Multi Layer Perception augmentation (SH+SIREN) can underperform without carefully designed objectives. Finally, gradient-boosted tree heads (XGBoost/LightGBM) most effectively exploit nonlinear interactions in the fused visual-geographic representation. These findings provide a scalable and principled recipe for improving satellite-based socioeconomic predictions using only publicly accessible data.