Aerial view crowd counting method based on reinforcement learning image brightness adjustment

The invention discloses an aerial view crowd counting method based on reinforcement learning image brightness adjustment. The method comprises the steps: firstly obtaining a labeled training set, and carrying out the preprocessing of the labeled training set, and obtaining a labeled data set of original-resolution image blocks and low-resolution image blocks; training based on the obtained data set to obtain an original crowd counting model and a rough crowd counting model; and then constructing and training an enhancement strategy network model. In the working process, for a to-be-processed unmanned aerial vehicle real-time image, the enhancement strategy network model obtained through training is used for selecting image blocks with corresponding resolutions, brightness values are adjusted, the corresponding crowd counting model is called according to the selection result to obtain the crowd counting result of each image block, and the crowd counting result of the whole image is comprehensively obtained. Ac