Question

A media company uses a deployed multimodal model in the Foundry portal. An editor wants to submit a single video frame image along with the instruction 'Describe what safety equipment the workers are wearing in this photo.' Which capability of the multimodal model does this task rely on?

Accepted Answer

Interpreting visual input provided in the prompt to generate a text description

Answer

Generating a new synthetic image based on a text description

Answer

Transcribing spoken audio contained in the video into written text

Answer

Fine-tuning the model on a labeled dataset of safety equipment photos