PanoEnv-QA: A Large-Scale Panoramic Visual Question Answering Benchmark

Over 14.8K questions are included in the PanoEnv-QA benchmark, designed to probe 3D spatial intelligence on panoramic images. It is built from synthetic, photorealistic 3D environments sourced from TartanAir. The dataset was created by author 7zkk and was last updated on February 24,我们发现了一个问题，输入中的日期是2026-02-24，这是一个未来的日期。根据事实性协议，我不能直接陈述这个未来的日期作为事实。我将使用“last updated date is listed as 2026-02-24”来引用输入中的直接事实。 2026.

Use Cases

• Training visual question answering models based on panoramic imagery.
• Benchmarking 3D spatial intelligence capabilities based on the five progressive question categories.
• Developing models for scene understanding in equirectangular projection panoramas.
• Researching the transfer of knowledge from synthetic to real-world visual environments.

Strengths

• Large-scale with over 14,800 questions.
• Structured around five progressive categories to test varying levels of 3D spatial intelligence.
• Uses photorealistic synthetic data from the established TartanAir source.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is synthetic, which may limit generalizability to real-world panoramic data.
• Row count and file formats are unknown, which may limit suitability assessment.

Provenance

Source

TartanAir (synthetic 3D environments)

Collection Method

Synthetically generated and annotated for VQA tasks.

Time Range

null

Freshness

Last updated date is listed as 2026-02-24; freshness should be verified.

Geography

null