CITYREP: A Unified Benchmark for Urban Representations Across Cities, Tasks, and Modalities

2026-05-25

AI Summary

Existing evaluations of urban representations are limited by a small number of cities and tasks, and random data splits often induce spatial leakage, hindering cross-regional generalization and fair model comparison. This work proposes CityRep—the first generalization-aware unified benchmark for urban representation learning—featuring a spatial-unit-agnostic evaluation framework and a block-based structured partitioning protocol. It establishes a multimodal, extensible suite spanning eight cities and eight diverse tasks. Through standardized alignment modules, block-level splitting strategies, and a reproducible evaluation pipeline, the study systematically assesses 11 representative models, revealing that random splits substantially overestimate performance and distort model rankings. The results underscore the necessity and effectiveness of CityRep in enabling rigorous cross-domain comparisons and aligning heterogeneous urban representations.

Abstract

Urban representation learning encodes complex urban environments into general-purpose embeddings for diverse downstream tasks and emerging urban foundation models. However, current evaluations are limited, typically focusing on one or two cities and tasks and relying on random splits that introduce spatial leakage, leading to inflated performance and weak support for cross-location generalization and fair comparison. To address this, we propose CityRep, a unified benchmark that evaluates urban representations across data modalities, cities, and tasks using spatially structured splits. CityRep consists of three key components: (1) a spatial unit-agnostic evaluation framework that supports heterogeneous urban representations through a standardized alignment module; (2) a unified evaluation protocol using block-based spatial splits to mitigate spatial leakage and enable rigorous model comparison; and (3) an extensible multi-city, multi-task benchmark suite spanning 8 cities and 8 tasks across regression, classification, and distribution prediction. We evaluate 11 representative urban representation models. Results show that performance is highly sensitive to the split protocol, with random splits inflating scores and altering model rankings. We also observe substantial variability across cities and tasks, underscoring the need for generalization-aware evaluation. CityRep is released as a reproducible benchmark with datasets, evaluation pipelines, and diagnostic tools to facilitate fair comparison and support future research in urban representation learning towards urban foundation models.

Problem

• urban representation learning
• spatial leakage
• cross-city generalization
• benchmark evaluation
• fair model comparison

Innovation

• urban representation learning
• spatial leakage
• cross-city generalization
• unified benchmark
• spatially structured splits