VQA

Visual Question Answering (VQA) is an AI technology that combines computer vision and natural language processing to generate precise text answers about the contents of an image.

VQA bridges the gap between sight and language by analyzing a digital image alongside a user's natural language question (e.g., "Is the store on the left open?" or "How many cars are parked in the driveway?") to deliver an accurate, context-aware response. The system relies on deep learning architectures (typically pairing a vision transformer like CLIP with a text decoder) to process visual features and linguistic cues simultaneously. This multi-modal capability powers critical real-world applications: it assists visually impaired users navigating daily environments, automates medical diagnostic checks on radiology scans, and streamlines surveillance analysis for security teams.

https://visualqa.org