Small target detection algorithm based on pyolo dynamic self-adaption

The invention relates to the technical field of target detection algorithms, and discloses a pyolo dynamic self-adaption-based small target detection algorithm, which comprises the following steps of: acquiring image data, performing data preprocessing, and dividing a data set into a training set and a test set by taking the data requirement of a YOLO model as a standard; performing optimization processing on the hyper-parameters in the training set by adopting a preset optimization search algorithm to obtain optimal parameters; wherein the hyper-parameters comprise the image size and the batch size; and training a preset pyolo target detection model by adopting the optimal parameters, and analyzing and verifying a result. According to the small target detection algorithm based on pyolo dynamic self-adaption, a search optimization algorithm is introduced into a YOLO structure, a unique pyolo target detection model is provided, self-adaption training of different types of data sets and target sizes is achieved