Visual Question Answering Subset with 30,000 Images

Encompassing 30,000 images from the GQA dataset, intended for training Visual Question Answering models. It is tagged for scene understanding and computer vision tasks, with associated English text.

Use Cases

• Train a Visual Question Answering model using the 30,000 images and associated English text for scene understanding.
• Fine-tune a vision-language model on the image and text pairs for tasks like image captioning or visual reasoning.
• Benchmark scene understanding algorithms using the provided image and text data from the GQA subset.

Strengths

• Contains 30,000 images for model training.
• Includes English text data for vision-language tasks.
• Specifically curated for Visual Question Answering applications.

Limitations

• The dataset is a subset, potentially limiting coverage of the full GQA data distribution.
• No column or sample data is provided to verify structure or content.
• The size, format, and specific image-text pairing details are unknown.

Provenance

Source

GQA Dataset

Collection Method

Subset extraction for VQA training.