Apparatus and method for visual localization with compact implicit map representation

The invention relates to a method of locating a mobile device in a region of interest, the method comprising the steps of: capturing an image using a camera of the mobile device, the camera having a current camera pose when capturing the image; determining, using a pre-trained image encoder of the mobile device, an image signature based on the image, the image signature being a representation of the current camera pose; iteration is performed comprising the steps of: selecting a pool of pose anchors from a map representation of the region of interest, each pose anchor corresponding to a candidate camera pose; generating a map signature for each pose anchor point, each map signature being a representation of the corresponding candidate camera pose; comparing the image signature with the generated map signature by determining a similarity score for each comparison; identifying a plurality of pose anchor points having the highest similarity score; wherein the initial iteration is performed based on an initial predefined pose anchor point pool, and in each subsequent iteration, the step of selecting the pose anchor point pool is based on the pose anchor points identified in the previous iteration; the current camera pose is estimated based on the pose anchor point identified in the iteration. The invention also relates to a corresponding mobile device.