Fiducial tags have attracted increased attention and use in the robotics community due to their relative ease of detection and pose estimation. Previous methods to detect these tags search the image pixel spaces in their entirety. With the advent of higher resolution video such as 4K and increased camera frame rate, however, it has become increasingly challenging to obtain real time tag detection, especially using lower-end computing devices. We present a novel algorithm that accelerates the detection speed of these fiducial tags by forecasting the future trajectory of the camera, devising a prediction for the upcoming tag location in the camera frame, and significantly reducing the image search space corresponding to the prediction. Based on the data from several test trials, we tuned the algorithm parameters and compared it to the conventional full search method in multiple real-world test experiments using AprilTags. In addition to the quantitative results, we present a qualitative analysis on the optimal application scenarios for our new algorithm. Our method has demonstrated between a three to four times increase in detection speed and is robust enough to handle intermittent tag occlusion.