DemoCut

Abstract

Amateur instructional videos often show a single uninterrupted take of a recorded demonstration without any edits. While easy to produce, such videos are often too long as they include unnecessary or repetitive actions as well as mistakes. We introduce DemoCut, a semi-automatic video editing system that improves the quality of amateur instructional videos for physical tasks. DemoCut asks users to mark key moments in a recorded demonstration using a set of marker types derived from our formative study. Based on these markers, the system uses audio and video analysis to automatically organize the video into meaningful segments and apply appropriate video editing effects. To understand the effectiveness of DemoCut, we report a technical evaluation of seven video tutorials created with DemoCut. In a separate user evaluation, all eight participants successfully created a complete tutorial with a variety of video editing effects using our system.

Soloist: Generating Mixed-Initiative Tutorials from Existing Guitar Instructional Videos Through Audio Processing

“…Interactive video tutorials have been studied intensively for various domains, such as software learning [1,4,11,23,24,29,39,49,53] or physical activity [12,25,38]. Prior work has investigated both the facilitation for authors to generate interactive video tutorials [11,12,25,51] and for users to effectively follow the video instructions [1,24,53]. Although useful, video tutorials are hard to navigate [1,33,51].…”

“…Although useful, video tutorials are hard to navigate [1,33,51]. Prior work has leveraged video segmentation to split an existing video into conceptual chunks to help people search for information [1,11,12,51]. In terms of the segmentation method, most closely relevant to our work is Waken [4], which reverse engineers an input software tutorial video to recognize UI components using computer vision techniques.…”

“…We first use Spleeter's 2-stem audio separation to obtain the voice and instrument track. Then, we use a similar approach as in DemoCut [12] to label non-silent sections across the waveform by analysing the loudness of each window for both tracks.…”

Automatic Video Creation From a Web Page

“…Video can be summarized by constraints [48,4,6] or presented as a storyboard [23]. Editing tools can be tailored for specific domains, such as conversational video [40,22], interviews [5], physical tasks [14,57] for particular types of footage [25,58]. Following these efforts, our work aims to automate the creation process for a specific domain-digital marketing.…”

“…For branding, creators consider the visual design of each shot, which commonly involves colors, sizes, and graphical layout. While editing one single video is a time-consuming process that could be partially automated [14,25,40], producing multiple videos from the same theme can be even more challenging. Recent trends encourage creators to publish different video versions based on the viewing experiences [55].…”