摘要 |
Implementing an object segmentation scheme where the object segmentation is achieved by comparison of a camera image and a scene reference image of an observation scene, with a learning device for generating the scene reference image, the learning device being designed for the purpose of generating the scene reference image by evaluation of a medium-term observation and/or long-term observation of the observation scene: <SL> <LI>a) Which is devised over a period of longer than one day, preferentially longer than several days, in particular longer than one week; and/or <LI>b) which is devised over a period that includes several states of the observation scene. </SL> |