CASUALTY RECOGNITION METHOD AND SYSTEM BASED ON DEEP LEARNING IN CASUALTY GATHERING PLACE SCENE

摘要

A casualty recognition method based on a deep-learning neural network. The casualty recognition method comprises the following steps: S10, acquiring at least one casualty picture in a casualty gathering place environment by means of a depth camera, and performing gathering to form a site data set of casualty pictures; S20, by means of data augmentation, generating an additional casualty picture of a smaller size for an original casualty picture captured at close range in the site data set, and associating the additional casualty picture with the original casualty picture and then storing the additional casualty picture in the site data set; and S30, inputting a site picture captured by the depth camera into a deep-learning-based neural network, so as to calculate and output the number of casualties in the site picture, wherein the neural network is trained using a pre-training data set and the site data set. Further provided is a casualty recognition system based on a deep-learning neural network. The system comprises a depth camera, a memory, and a processor, wherein the processor implements the method when executing instructions stored in the memory.