Target identification method based on Farneback optical flow method and time sequence neural network and related equipment

• Main Authors: LIU XI, ZHOU MEIJUAN, WANG NING, MAO YUANHONG, HE PENGCHAO
• Format: Patent
• Language: Chinese; English
• Published: 05.03.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a target recognition method based on a Farneback optical flow method and a time sequence neural network and related equipment, and belongs to the field of image target recognized.The method comprises the steps that real category marking is conducted on a training image and a test image, and a training sample set and a test sample set are obtained; constructing a tiny target recognition network which jointly utilizes target space information and time sequence information, inputting the training sample set into the tiny target recognition network, and training a network model by using an optimization algorithm; after the model is converged, inputting the test sample set into the trained network model to obtain a final recognition result; according to the method, the spatial information of a single target is not used, the relevance between different target characteristics and image characteristics is utilized, the time sequence information of multiple frames of images is adopted for targe