Benchmark RadarAI BENCHMARK PROFILE
FormStruct-Bench
The paper introduces FormStruct-Bench, an evaluation dataset for table-form document structure recognition, but no official links to the benchmark artifacts are provided.
- Released
- 2026-08-11
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
An evaluation gap exists between holistic document outputs and fine-grained structural scores, but the lack of public benchmark access limits its practical utility.
Motivation
Transforming table-form documents into machine-processable records requires recovering not only their visible content but also the multilevel structure that organizes it.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.