Network architecture search method for YOLOv5 target identification based on multi-target genetic algorithm

The invention provides a network architecture search method for YOLOv5 target identification based on a multi-target genetic algorithm. The method comprises the following steps: determining a searchable space of a YOLOv5 feature extraction backbone network; based on a high-precision coding mode of various candidate operations in a searchable space, determining a corresponding relation between different network models and coding and decoding in the YOLOv5 feature extraction backbone network; determining a plurality of target functions selected in the multi-target genetic algorithm based on the network model parameters; encoding the plurality of initialized network models to form a network model population, and training and verifying on a limited data set after decoding to obtain numerical indexes of a plurality of objective functions corresponding to the network models under the encoding; according to the network model codes and the numerical indexes of the multiple corresponding objective functions, a multi-o