Potted flower intelligent identification method based on deep learning

The invention relates to the field of flower identification, in particular to a pot flower intelligent identification method based on deep learning, and the method comprises the steps: constructing a pot flower detection data set which comprises a plurality of labeled pot flower images; a YOLOv5sNMRH target detection model is constructed, the YOLOv5sNMRH target detection model is based on YOLOv5, an NAMMobileNetV3s backbone feature extraction network is adopted by a Backbone part, a bilinear interpolation method is adopted by up-sampling of a Neck part, and a decoupling detection head Resdecoupled Head added with a residual edge is adopted by a Head part; the potted flower detection data set is adopted to train the YOLOv5sNMRH target detection model; and utilizing the trained YOLOv5sNMRH target detection model to identify the potted flower in the image of the potted flower to be identified. According to the invention, the types of potted flowers can be accurately identified. 本发明涉及花卉识别领域，具体涉及一种基于深度学习的盆栽花卉智能识别方