Abstract: The unique geometric morphology of harbor seal whiskers enables exceptional underwater target recognition and tracking. This biomimetic prototype offers significant potential for novel sensor development. A seal whisker-inspired sensor was fabricated with single-degree-of-freedom swing capability. Flume experiments tested three background flow velocities and seven upstream cylinder types. Angular displacement data were recorded for the whisker model across all test conditions. Convolutional Neural Networks (CNN) trained on this displacement data achieved accurate shape identification for most upstream cylinders. Adjusting training sample length and setting low-pass filter cutoff frequency during preprocessing impacted recognition results. Increasing sample length improved performance, peaking at length 300. Further length increases provided negligible gains. Low-pass filter cutoff frequencies beyond 10 Hz caused no significant change in effectiveness, confirming that useful angular displacement signal components reside primarily below 10 Hz. Development of whisker-inspired sensors using angular displacement signals proves feasible.