Video summarization by group scoring

Abstract

Abstract-In this paper a new model for user-centered video summarization is presented. Involvement of more than one expert in generating the final video summary should be regarded as the main use case for this algorithm. This approach consists of three major steps. First, the video frames are scored by a group of operators. Next, these assigned scores are averaged to produce a singular value for each frame and, lastly, the highest scored video frames alongside the corresponding audio and textual contents are extracted to be inserted into the summary. The effectiveness of this approach has been evaluated by comparing the video summaries generated by this system against the results from a number of automatic summarization tools that use different modalities for abstraction.

Personalized video summarization based on group scoring

“…SCORING: In previous work [10], we have described an approach to video summarization based on a group scoring method, in which original video frames are scored by a number of video scorers (experts) and the assigned scores averaged to produce a singular value for each frame. A group of frames with the highest average scores are then chosen to be inserted into the final summary.…”

“…In accordance to the summarization method based on group scoring, [10] the highest scored frames alongside the audio and textual content are selected and inserted into the final video digest. Considering the required number of frames, those highest scored frames will be selected to be added to a final list and to be sorted based on their time order in the original video.…”

“…Moreover, no information regarding the corresponding adopted summarization tools for each of the summary versions was revealed to participants. After watching the original video and the summaries the users were asked to score each of the generated abstracts awarding marks between 0 (worst video summary possible) to 10 (best video summary possible), from 4 different perspectives consisting of Recall (Re), Precision (Pe), Timing (Ti) and Overall Satisfaction (OS).These measures were described in details in our previous work [10]. The given scores for each of these measures were averaged over 30 users and their mean values for each of the video categories are given in Table I.…”

“…The given scores for each of these measures were averaged over 30 users and their mean values for each of the video categories are given in Table I. S1 [12], S2 [13], S3 [14], S4 [10] and S5 indicate the average achieved scores by, respectively, the first, second, third, fourth and our recent proposed personalized systems…”

Personalized video summarization by highest quality frames

“…In a proposed model in [16], a user-centered approach to video summarization based on a group scoring was suggested, in which original video frames are scored by a number of video scorers (experts) and the assigned scores averaged to produce a singular value for each frame. A group of frames with the highest average scores are then chosen to be inserted into the final summary.…”