Method, computer program, storage medium, person detector and monitoring arrangement for person detection

A non-transitory, machine-readable storage medium comprising instructions that when executed by a computer cause the computer to:

• monitor a monitoring area (1) via a plurality of cameras (2a, b, c, d), wherein the cameras (2a, b, c, d) each provide image data for a respective monitoring area section (3a, b, c, d), and
• switchingly operate a person detector in a detection mode and in an adaptation mode, wherein the person detector comprises a machine learning algorithm based on an original model for analyzing the image data and for detecting persons in the monitoring area (1), wherein, for the adaptation mode, the monitoring area (1) is temporarily restricted to a limited monitoring area (4), wherein,
• the image data from the limited monitoring area (4) are recorded, wherein a domain adaptation of the original model to an application model is carried out based on the image data of the limited monitoring area (4), wherein the limited monitoring area (4) is a subset of the monitoring area (1), wherein the monitoring area (1) is restricted in order to specify a target trajectory (5) for persons to the limited monitoring area (4) and the target trajectory (5) includes and/or defines points of change, wherein a change of direction and/or a pivot of view is specified at the points of change of a person, wherein,
• in the detection mode, the machine learning algorithm adapted to the application model is used to analyze the image data, for person detection and/or for person tracking.