Benchmark RadarAI BENCHMARK PROFILE
TC-Bench
A benchmark dataset for tropical cyclone research with an automated construction pipeline, used to probe scientific alignment of vision foundation models.
- Released
- 2026-05-23
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Examines whether models rely on visual shortcuts rather than structural invariants in scientific domains.
Motivation
While Vision Foundation Models (VFMs) excel at predictive tasks on satellite imagery, their performance can arise from visual correlations rather than underlying structural invariants, making even perception-based out-of-distribution accuracy a poor proxy for scientific utility.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.