Field seedling stage grass identification method and device

Abstract

The invention provides a field seedling stage grass identification method and device. The method comprises the following steps of inputting an acquired field seedling stage grass image into a field seedling stage grass identification model; and outputting a field seedling stage grass type corresponding to the field seedling stage grass type image, with the field seedling stage grass type identification model being obtained by migrating the initially trained InceptionV3 model to a field seedling stage grass type sample image set for training. The InceptionV3 model is directly migrated to a field seedling stage grass sample image set for training after initial training. As the InceptionV3 model has a certain image recognition capability after initial training, the InceptionV3 model is migrated to the field seedling stage grass sample image set for training, the training efficiency can be improved, and the training efficiency and recognition accuracy of the field seedling stage grass recognition model can be improved.

Description

technical field [0001] The invention relates to the technical field of deep learning and image processing, and in particular to a method and device for identifying grasses at the seedling stage in the field. Background technique [0002] Field weeds are very harmful to crop seedlings because they compete with crops for sunlight and nutrients. Generally, the growth of weeds in the field is controlled by applying pesticides to the weeds in the field. However, there are various types of weeds in the field, and the weeds in the field are very similar to the seedlings of crops at the seedling stage. Regular pesticide spraying will cause waste of resources and environmental pollution, and excessive pesticides will endanger human health. Therefore, the rapid and accurate identification of field grasses is of great significance for the removal of field weeds. [0003] Traditional methods for identifying weeds include manual identification and remote sensing identification. Among t...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the problem that the currently used machine vision recognition method has poor recognition effect, or the structure of the network model is complex, and a large amount of data is required for training, resulting in low efficiency, the embodiment of the present invention provides a method and device for field grass recognition in the seedling stage

Images