Words, videos, and popes

Motion Words for Videos

In the task of activity recognition in videos, computing the video representation often involves pooling feature vectors over spatially local neighborhoods. The pooling is done over the entire video, over coarse spatio-temporal pyramids, or over pre-determined rigid cuboids. Similarly to pooling image features over superpixels in images, it is natural to consider pooling spatio-temporal feature...

Can Toddlers Learn Novel Words from Educational Videos

Topic Transition in Educational Videos Using Visually Salient Words

In this paper, we propose a visual saliency algorithm for automatically finding the topic transition points in an educational video. First, we propose a method for assigning a saliency score to each word extracted from an educational video. We design several mid-level features that are indicative of visual saliency. The optimal feature combination strategy is learnt from a Rank-SVM to obtain an...

Learning to Lip Read Words by Watching Videos

Our aim is to recognise the words being spoken by a talking face, given only the video but not the audio. Existing works in this area have focussed on trying to recognise a small number of utterances in controlled environments (e.g. digits and alphabets), partially due to the shortage of suitable datasets. We make three novel contributions: first, we develop a pipeline for fully automated data ...

What's News, What's Not? Associating News Videos with Words

Text retrieval from broadcast news video is unsatisfactory, because a transcript word frequently does not directly ‘describe’ the shot when it was spoken. Extending the retrieved region to a window around the matching keyword provides better recall, but low precision. We improve on text retrieval using the following approach: First we segment the visual stream into coherent story-like units, us...