Mature apple identification method based on full convolutional neural network

Abstract

The present invention discloses a mature apple identification method based on a full convolutional neural network. The specific steps of the method of the invention are as follows: Step 1, collectingapple pictures and annotating mature apples to make a training data set; Step 2, inputting pictures and carrying out noise reduction processing; Step3, designing a structure of a full convolutional neural network; Step4, training the full convolutional neural network by using the noise reduction-processed training data set; and Step5, using the trained convolutional neural network to carry out identification and segmentation on the mature apples. According to the method disclosed by the present invention, the image is segmented by using a full convolutional neural network, and the prediction category of each pixel of the output image can be obtained, so that an accurate position of the apple can be obtained, and the problem of inaccurate identification accuracy of the apple under the condition that the background is too complicated is avoided; and the method has high accuracy, efficiency and real-time performance, and the implementation process is convenient and the general-purpose technology is strong.

Description

Technical field [0001] The invention relates to a mature apple recognition method based on a full convolutional neural network, which belongs to the technical field of digital image processing target detection and recognition. Background technique [0002] my country’s fruit output has always been the number one in the world, and the fruit planting area is increasing year by year. However, in this era of rapid development of fruit planting area, the form of apple picking in my country is still very backward or manual picking is the main method. This kind of picking method is not only low in efficiency and low in quality, but also inconsistent with the development trend of agricultural modernization. For example, since a person can only pick apples from one tree at a time, it greatly limits the person's picking efficiency. Failure to pick ripe apples in time will result in overripe or even rot. For this reason, robots have played a very important role in agricultural picking. How...

AI Technical Summary

Problems solved by technology

Failure to pick ripe apples in time will lead to overripe or even rotten apples

For this reason, robots have played a very important role in agricultural picking work, but at this stage, the main problems faced by domestic and foreign agricultural picking robots are: low recognition accuracy, low efficiency and poor applicability

Images