35 upvotes
CodeCap-Bench asks if code can replace video captions
What is CodeCap-Bench: Can Code Serve as a Good Video Caption??
The paper presents CodeCap-Bench, a benchmark that evaluates whether programming code can serve as a concise, precise caption for videos. It aims to capture fine-grained spatial and temporal details that natural-language captions often miss.
Who it is for
Researchers building video understanding or captioning systems should look at it to test and improve models that output structured code instead of text.