A method for detecting ripeness of bunched tomatoes based on deep learning and computer vision

Abstract

The present invention relates to the field of computer vision technology and maturity detection technology of string tomato, in particular to a method for detecting maturity of string tomato based on deep learning and computer vision. The method includes: establishing a first-level SSD target detection model and a second-level AlexNet target detection model based on deep learning; obtaining the position and confidence information of all detected targets output by the last layer of the first-level SSD target detection network First-level area information; calculate the actual length of each detection target in the world coordinate system in the image to be detected, and judge whether the actual length meets the qualification conditions for string tomatoes; obtain the output of the last layer of the second-level AlexNet target detection network containing the Detect the location of all individual fruits of the target and the secondary area information of the confidence information; calculate the ripeness of the bunch tomato fruit. The invention has the advantages of fast identification and detection speed, strong generalization ability and strong portability, and realizes real-time detection of ripeness of string tomato fruit.

Description

technical field [0001] The present invention relates to the field of computer vision technology and maturity detection technology of string tomato, in particular to a method for detecting maturity of string tomato based on deep learning and computer vision. Background technique [0002] String tomato, also known as ear tomato, is a tomato variety that is harvested in clusters and marketed. Its outstanding advantages are good fruit quality, thick flesh, rich in multivitamins, special taste, easy planting and good planting benefits. Therefore, Greenhouse string tomato has become a vegetable variety that the majority of vegetable farmers are vying to plant. At present, the string tomatoes grown in my country are basically harvested manually, and the picking cost is about 10,500 yuan / hm 2 , accounts for more than 30% of the total production cost, and the use of automated harvesting is of great significance to ensure the safe supply and efficient production of string tomatoes. ...

AI Technical Summary

Problems solved by technology

[0003] In recent years, many tomato detection and recognition algorithms based on traditional algorithms have been proposed one after another. Due to the limitations of traditional algorithms, traditional algorithms have poor scene adaptability, and the detection effect is relatively poor for situations where the illumination changes significantly, the background interference is serious, and the target is blocked. Therefore, in order to improve the recognition accuracy of bunch tomato fruits in an unstructured environment and improve the efficiency of automatic picking of high-quality ripe bunch tomatoes, the traditional algorithm is gradually replaced by a more robust deep learning detection algorithm.

Images