Twin network tracking system and method based on space-time attention mechanism

• Main Authors: LIU RUHAO, LU XIANLING
• Format: Patent
• Language: Chinese; English
• Published: 05.07.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a twin network tracking system and method based on a space-time attention mechanism in the technical field of target tracking of computer vision. The method comprises the following steps: acquiring template image data and search image data in image data; respectively inputting the template image data and the search image data into a template branch and a search branch for feature extraction to obtain template feature data and search feature data; inputting the search feature data into a space-time attention mechanism module to obtain space-time attention information features; performing cross-correlation on the template feature data and the space-time attention information features to obtain a response diagram; and inputting the response diagram into a classification and regression network for pixel-by-pixel classification and regression, classifying each pixel point, and calculating the distance from the pixel point to the prediction box to obtain a classification result and a boundin