Feature point detection model training method

• Main Authors: LI QIDONG, QI ZIMING, LI ZHIYANG, ZHANG WEI, FU SONGLIN
• Format: Patent
• Language: Chinese; English
• Published: 27.09.2019
• 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 feature point detection model training method which is suitable for being executed in computing equipment and comprises the steps of acquiring a training image with annotation data, wherein the annotation data is feature point coordinates of a target object in the image; inputting the training image into a first model for processing to obtain a trained first model; processing the trained network structure of the first model to obtain a second model; initializing network parameters of a second model based on the trained network parameters of the first model to obtain an initialized second model; and inputting the training image into the initialized second model for processing to output predicted feature point coordinates of the target object in the training image, and training the initialized second model by using the predicted feature point coordinates and the marked feature point coordinates to obtain a trained second model as a feature point detection model. 本发明公开了一种特征点检测模型的训练方法，适于