AgentVQA: Multi-Domain Visual Question Answering for Agents

AgentVQA is a multi-domain dataset for training and evaluating visual agents. The dataset contains multiple-choice questions based on screenshots, images, and videos across five domains, including GUI interaction and robot manipulation. It was created by AgentVQA and last updated on December 18, 2025.

Use Cases

• Train visual agents for GUI interaction based on screenshots.
• Evaluate spatial reasoning capabilities of AI models based on images.
• Benchmark game understanding agents using visual question answering.
• Develop robot manipulation policies based on visual perception tasks.
• Test video perception models with multiple-choice questions.

Strengths

• Covers five distinct domains including GUI interaction and video perception.
• Last updated on December 18, 2025, indicating recent maintenance.

Limitations

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

Provenance

Source

AgentVQA

Freshness

Last updated 2025-12-18 16:16:47