In radar target detection, Constant False Alarm Rate is a commonly employed detector known for its simplicity and effectiveness. Its sliding-window detection mechanism possesses computational similarity to convolutional operations in machine learning. With the increasing emergence of AI-enhanced radar processing algorithms, systems at the edge tend to be equipped with machine learning accelerators to expedite matrix multiplications and convolutions. This paper introduces a heuristic algorithm that equivalently transforms Cell-Averaging Constant False Alarm Rate (CA-CFAR) to a convolutional operation coupled with nonlinearity. Comparative experiments of the transformed CA-CFAR utilizing machine learning accelerators against the classical CA-CFAR executed on ARM cores demonstrate that our proposed method significantly reduces the processing latency by 35 to 47 times and saves energy by 30 to 40 times. This advancement brings substantial promise for facilitating real-time high-resolution radar target detection without dedicated CFAR accelerators.