Positioning, Navigation and Timing (PNT) methods in GNSS-challenged environments require multi-sensor and cooperative approaches to mitigate low or complete unavailability of GNSS measurements. Many methods also rely on map databases and the availability of sensors throughout the environment. Data like Signal of Opportunity (SoO) ranges, Inertial Measurement Units, and camera data are often used to ensure measurement redundancy. Given the recent advancements in Artificial Intelligence (AI) image segmentation, especially Meta's Segment Anything Model (SAM), there is an opportunity to treat AI as a modern SoO. SAM can quickly and efficiently recognise distinct objects in any image while the Depth Anything (DA) model can create a pixel-based depth map from any image. A novel architecture for combining multi-sensor cooperative positioning and position integrity method with SAM and DA is proposed. In the proposed architecture, SAM features of interest will be used to improve the existing Spatial Feature Constraint algorithm to constrain vehicles to road features and to determine ranges to infrastructure from segmented images using the DA model. In this paper, the initial feasibility study of using SAM and DA to determine ranges from images is carried out. SAM and DA are tested on photographs taken in Vienna, Austria. The feasibility of establishing a functional relation between determined depth and ground truth distances is studied and feasibility is demonstrated.