Imitate Before Detect is an innovative text detection technology aimed at enhancing the capability to detect machine-revised text. By imitating the stylistic preferences of large language models (LLMs), this technique more accurately identifies text modified by machines. Its core advantage lies in effectively distinguishing subtle differences between machine-generated and human-written content, thereby holding significant application value in the field of text detection. The background information indicates a substantial increase in detection accuracy, achieving a 13% improvement in AUC when handling open-source LLM revised texts, and enhancements of 5% and 19% in detecting GPT-3.5 and GPT-4 revision texts, respectively. It is positioned as an efficient text detection tool for researchers and developers.