A series of 24 short, 30-second video segments captioned at different speeds were shown to 578 people. The subjects used a five-point scale (Too Fast, Fast, OK, Slow, Too Slow) to make an assessment of each segment’s caption speed. The “OK ” speed, defined as the speed at which “Caption speed

captions that were submitted to a contest hosted by the New Yorker. Having access to thousands of cap-tions submitted for the same image allows us to analyze the breadth of responses of people to the same visual stimulus. We first describe how we acquire judgments about the humorousness of different

The caption package was superseeded by the new caption2 package. So please use caption2 instead; for migrating please see the enclosed manual.

This paper presents a novel approach for automatically generating image descriptions: visual detectors and language models learn directly from a dataset of image captions. We use Multiple Instance Learning to train visual detectors for words that commonly occur in captions, including many different

ideo segments captioned at different speeds were shown to a group of 578 people that included deaf, hard of hearing, and hearing viewers. Participants used a five-point scale to assess each segment's caption speed. The "OK " speed, defined as the rate at which "caption speed

ideo segments captioned at different speeds were shown to a group of 578 people that included deaf, hard of hearing, and hearing viewers. Participants used a five-point scale to assess each segment's caption speed. The "OK " speed, defined as the rate at which "caption speed

In this paper, we present CaptionEye/KE, a Korean to English machine translation system that is applied to a practical TV caption translation. And its experimental evaluation is performed on actual TV news caption texts that are extracted by caption extractor. This system adopts a highly robust HMM

in captions, including many different parts of speech such as nouns, verbs, and adjectives. The word detector outputs serve as conditional inputs to a maximum-entropy language model. The language model learns from a set of over 400,000 image descriptions to capture the statistics of word usage. We capture

We have developed a novel system for baseball video event detection and summarization using superimposed caption text detection and recognition. The system detects different types of semantic level events in baseball video including scoring and last pitch of each batter. The system has two

, meetings etc.) The provision of synchronized text captions (subtitles) with video enables all their different communication qualities and strengths to be available as appropriate for different contexts, content, tasks, learning styles, learning preferences and learning differences. For example, text can