In this study, distributed shape reconstruction is demonstrated based on a ring-core few-mode fiber (RC-FMF) and differential pulse-pair Brillouin optical time-domain analysis (DPP-BOTDA). The RC-FMF is adhered to a thin steel substrate, such that curvature can be characterized by the bending-induced strain, and the curvature-direction can be distinguished by identifying the compressive or tensile strain-induced down- or up-shift of the Brillouin frequency of the proposed sensor. An asymmetric hump shape is successfully reconstructed via the Frenet-Serret-frames algorithm. The RC-FMF exhibits bending-induced loss as low as 0.02 dB when the bending radius is reduced down to 0.285 cm, which showed the advantage of large curvature shape sensing over standard single-mode fiber.