| dc.description.abstract | Biometrics reduce lost identification items, such as identity cards or employee cards... and forget your account number or password. In recent years, biometrics have been continuously studied, and many are widely used in life, such as: face recognition, fingerprint unlocking, iris identification, palm recognition... And so on, this article focuses on palm print-related research. Compared with the traditional palm print identification methods are mostly contact-type identification or palm print, will cause users inconvenience and practicality. In order to improve this problem, this study proposed a non-contact palm detection system, first with a smartphone at different times and places to take images, and then use label AppleImg to mark the area of interest in palm print, and then YOLOV3 and YOLOX convolutional neural network model training, after training with the mAP model evaluation index, YOLOX 1.0, YOLOV3 for 0.995. This study designed palm grain detection, in a small number of learning samples, but also has a good model evaluation results, and can be detected in different environments palmprint area detection. | en_US |