Accurately determining the shape and tip position of puncture needles in surgery is challenging due to the limitations of traditional sensors. This work introduces a Fiber Bragg Grating (FBG) sensor designed for this purpose. The sensor bonding and calibration platforms were developed, and a shape perception algorithm based on differential geometry was implemented. For a 150 mm puncture needle, the average tip positional error was 3.687 mm, with a maximum error of 6.832 mm. Additionally, a flexible hose-based curvature sensor was created for real-time sensing of the needle tip’s curvature. For a 30 mm needle, this sensor achieved an average positional error of 0.244 mm and a maximum error of 0.528 mm.