Content-based video sequence editing: a client-server Java system for user-assisted segmentation

In this paper we present a semiautomatic video sequence segmentation technique, where the classification of the pixels of a video sequence into two classes, (moving and still) exploits parameter tuning and region marking provided by the user. User assistance is a straightforward way to increase segmentation performance, enabling the user to add high-level information in the process. The semiautomatic system is based on a client-server architecture that allows the design of a multi platform user interface by Java language and makes the system accessible on the network. This architecture allows the segmentation process to run on high speed servers, and it is configurable to give a segmentation service on the network to users who can upload their sequences to segment and download the resulting segmentation masks.