Conceptdense captioning
Dense Captioning
A technique for generating natural language descriptions of video frames or segments that cover elements such as objects, scenes, and actions. A common approach calls a vision-language model every few frames to generate text as historical memory. Its advantage is compactness and direct usability by language models; its cost is being highly lossy and lacking temporal resolution.
Timeline (last 90 days)
Oct 15
The dense description method generates text descriptions as historical memory by calling a vision-language model every few frames, at the cost of being highly lossy, often glossing over the temporal transitions between actions and the trigger conditions for state switches.
Unverified50%