ParaDLC-Bench: Multi-Region Localized Image Captioning Benchmark

ParaDLC-Bench (Parallel Detailed Localized Captioning Benchmark) is a benchmark for multi-region localized captioning created by MSALab. It extends DLC-Bench from single-region to concurrent multi-region evaluation, stressing a model's ability to describe many regions at once without cross-region interference. The dataset was last updated on June 19, 2026.

Use Cases

• Benchmarking caption quality for models that generate descriptions for multiple image regions simultaneously.
• Evaluating inference efficiency of multi-region captioning models.
• Training models to avoid cross-region interference when describing concurrent regions.
• Researching localized captioning techniques that extend beyond single-region evaluation.

Strengths

• Explicitly designed for concurrent multi-region evaluation, extending a prior benchmark.
• Jointly evaluates both caption quality and inference efficiency, a key feature mentioned in the description.
• Created by MSALab, a known research entity, and associated with a research paper and code.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and data size are unknown, which may limit suitability assessment.
• License information is unknown, which could restrict usage.

Provenance

Source

MSALab

Collection Method

Likely constructed as a benchmark for research, extending the DLC-Bench dataset.

Freshness

Last updated 2026-06-19 03:59:07; freshness should be verified.