On Eye

OnEye

Project Manager: Prof. Dr. Didier Stricker
Funding by: BMBF

Digital video is becoming a dominant element of commercial web-sites and is definitively the driving force behind the current expansion of the web and the new “Generation Mobile”.The presentation of a product with help of a video increases dramatically the shopping experience of the online-customer.

Digital video is becoming a dominant element of commercial web-sites and is definitively the driving force behind the current expansion of the web and the new “Generation Mobile”. The presentation of a product with help of a video increases dramatically the shopping experience of the online-customer. Video is becoming an essential medium against classic product description with text and pictures. Moreover the apparition of new end-devices such as large-screen smartphones and web-pads (iPad, webTab) as well as novel content platforms (Youtube, WebTVs) made video to a standard medium for Internet communication.

The next development step will consist in extending videos to embedded and local interaction possibilities. Objects or persons in a video represent interactive “tags”, which can be selected and hereby provide additional information. A promising commercial application of this technology is “Embedded Advertisement”. Banners, product information or hyperlinks to an e-shop are attached to a given product in the video. The user watches a movie or TV-show naturally, without any product hints or additional messages as currently blended in most of the videos portals. She/He has however the possibility to move his/her PC-mouse or remote control over the video-viewer and to let appear the tags marking the available products. He/she can then stop the video, and click on the marked objects and get wished information, or go directly to the online-shop.

While the fundamental concepts of “interactive-videos” are well-known and all the required mechanisms are available in the major video-players, their production is still very limited, since no reliable tool for automatic object tracking exists on the market. The goal of this project is therefore to develop an effective video-editor for the easy post-production of interactive videos, relying on image processing, online learning methods and adapted user input. The algorithms should enable to track an object over long sequences in a reliable way in spite of appearance changes and partial occlusions. The user input will be designed in order to support the tracking process and thus enable effective and fast creation of interactive videos.