Plant recognition method based on deep learning and plant classification

Abstract

The invention discloses a plant recognition method based on deep learning and plant taxonomy. The sample plant image is marked by the subject, the genus and the species; the sample plant image is input into a deep convolutional neural network for training. And the loss function are set as the weighted sum of the cross entropy loss of the subject, the genus and the type of tags, the weight of the neural network is updated through a random gradient descent algorithm, the training concludes after the deep convolutional neural network converges, and the weight of each layer without changing is fixed to obtain the trained deep convolutional neural network. According to the invention, deep learning is combined with plant taxonomy, and a subject label is introduced as a learning target, so that the recognition accuracy is improved.

Description

technical field [0001] The invention relates to the technical field of plant identification, in particular to a plant identification method based on deep learning and plant taxonomy. Background technique [0002] Generally speaking, there are two types of methods in the field of image recognition. One is the recognition method based on traditional machine learning, which is characterized by the fact that the feature extraction method is generally manually defined; Convolutional Neural Network Automatic Extraction. However, the former often introduces artificial bias, and the latter has a certain degree of overfitting problem, which affects the recognition performance. [0003] The existing patent CN201810030151.X discloses a plant species identification method and device, which obtains pictures of at least two of the three parts of the flower, leaves and the whole plant of the plant to be identified as pictures to be identified; The saved multiple recognition models respec...

AI Technical Summary

Problems solved by technology

However, the former often introduces artificial bias, and the latter has a certain degree of overfitting problem, which affects the recognition performance.

Images