Benchmark RadarExpoMotion
ExpoMotion is a large-scale benchmark for multi-exposure fusion with dynamic scenes, containing 1,738 sequences and 10,909 images across diverse environments. It provides high-fidelity ground truth for reference-based evaluation and a separate set for no-reference evaluation. The benchmark includes training and testing splits with controlled and real-world motions.
- Released
- 2026-07-03
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
Existing multi-exposure fusion benchmarks often neglect dynamic scenes and lack reliable ground truth, hindering evaluation of deghosting capabilities. ExpoMotion addresses this gap by providing a large-scale dataset with high-quality ground truth, enabling reproducible comparison of methods that handle motion-induced artifacts. This supports practical deployment in real-world scenarios where motion is common.
Motivation
Multi-Exposure Fusion (MEF) effectively extends dynamic range, but practical deployment is hindered by motion-induced ghosting and the scarcity of high-quality dynamic benchmarks.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.