Detecting and recognizing text in a video is a challenging task. The task becomes harder in real-time video. The reason is that in real-time video we have to face the video in different lighting conditions. In low light or in highly bright lighting conditions the ambiguity in a video becomes high. Detecting text in such videos is difficult. Here we propose a model FLiTe (Fuzzy Logic Based Text Detection in Low Light) where we generate a fuzzy based inference system to detect and recognize the text from real-time video in low light conditions. The proposed method detects and recognizes the text in real-time with low computational time. The method is tested on various real-time low light videos. The performance of the method is found to be quite satisfactory both qualitatively and quantitatively.