Mama-based time sequence prediction unmanned aerial vehicle single target tracking method

Abstract

The invention discloses a Mama-based time sequence prediction unmanned aerial vehicle single target tracking method. The method specifically comprises the following steps: step 1, selecting a training set and a test set for target tracking; 2, preprocessing the training set and the test set selected in the step 1 to obtain a processed training set picture pair and a processed test set picture pair; 3, constructing a Mama-based time sequence prediction single-target tracking network model in an unmanned aerial vehicle scene, and obtaining a tracking result; 4, training the model constructed in the step 3 by adopting the training set picture pair preprocessed in the step 2, and outputting the trained model; and step 5, inputting the test set picture pair preprocessed in the step 2 into the model trained in the step 4, and finally obtaining a tracking target image with a detection frame. According to the invention, the problem of low tracking success rate when shielding, deformation and disappearance occur in a video frame in a traditional unmanned aerial vehicle single-target tracking method is solved.

Images