Logjam

Logjam is designed to run at the same time as video is being shot and captured. With Logjam, the videographer can create and modify the list of keywords at any time during the shoot. Logjam maintains the list of all video clips, which is updated automatically as new video is captured and saved. The videographer can choose to let Logjam automatically assign a desired subset of keywords to new clips as they are created, without manual review and logging. Each clip's individual set of keywords can be edited separately for fine-tuning.

Logjam frees the videographer from having to annotate each video clip manually. With Logjam, the set of descriptions can evolve during the production, rather than being defined before or after the production. This is beneficial because the videographer has an ongoing overview of the annotations, and could identify missing or newly desirable footage.

Background

Keyword annotations are a basic form of descriptive metadata for video content. Annotations can be used by applications to search a video database or to perform automated editing. Some general information about video can be gathered easily, such as duration of shot or time of shooting. Image analysis and object recognition techniques can yield information about what is in a shot. Still, the most useful annotations are likely to be those defined explicitly by the director / storyteller. He or she can define particular keyword categories and values, in order to describe the aspects of the content that are salient to the story and the narration. For example, the emotional tone of a shot may be crucial or inconsequential depending on what the story is and how it will be told.

These useful keyword annotations have to be entered manually, usually by reviewing all of the footage before beginning post-production - a time-consuming process. If annotations could be attached to video content during the shoot, it would save time later. More importantly, it would also help the videographer achieve fuller coverage during the shoot, since the set of descriptions can evolve with the captured source material. The evolving annotation set would make it easier to identify missing or newly desirable footage. Logjam is a modest illustration of how such a functionality might be provided to videographers who are shooting with computationally powerful "smart" cameras.

Implementation

Logjam currently runs on the Sony Vaio GT3-K, which combines the Vaio laptop and a video camera into one physical unit. The video is captured through Sony's Smart Capture application; both Logjam and Smart Capture run at the same time. This hardware / software combination illustrates how an enhanced "smart" camera running Logjam might facilitate the task of video annotation. Logjam does not depend on any specific feature of the GT3-K or Smart Capture, and it can run on other similar devices.

James Jung-Hoon Seo | 3.19.2002 | Home