Traditional PIV measurements only provide planar velocity data, and simultaneous spatially resolved measurements of temperature fields can only be performed with a different and more elaborated setup using temperature-sensitive tracer particles. PINNs have already shown great potential for improving the quality of PIV measurements [1] by combining data information and the knowledge of the governing equations. In this study, we present a method to infer temperature fields from planar PIV data in turbulent Rayleigh-Bénard convection (RBC), using a PINN approach that extends the methods outlined in [2, 3]. The physical setup is a cubic RBC cell (cf. fig. 1). With data only available in a plane A : x = x 0 , the residuals of the governing PDEs are solved in an enclosing 3D domain with thickness δ x . The parameter δ x must be carefully chosen within 0 < δ x < 1, depending on the flow normal to A. For developing several methodological approaches, we take advantage of available DNS data to mimic PIV measurements. Results with three velocity components show encouraging outcomes, with the coefficient of determination (R 2 ) exceeding 0.99. Reducing the input to just the two in-plane velocity components, requires the inclusion of additional physical constraints, due to the lack of information about the flow perpendicular to A. Our study provides a foundation for future research focused on inferring temperature fields through in-field measurements in industrial applications with more complex geometries.