The traditional method of measuring the lettuce height is a manual measurement with instruments, which is greatly affected by human error.At present, researchers have proposed to use color cameras to obtain RGB images of lettuce, and to obtain the height of lettuce from the images. However, these tasks usually require camera calibration or a reference object with a known height, which is somewhat restrictive. Considering that deep neural networks have a powerful ability to feature extraction and expression, without camera calibration and reference objects, we try to use four networks of image recognition to explore the effect of deep learning on abstracting the lettuce height from RGB images. On the test set, including 80 images and height from 0.9 cm to 7.5 cm, we achieve a good result with a mean absolute error of 1.22 mm.

To obtain phenotypic parameters by means of lossy measurement, we proposed a comprehensive and integrated approach to predict different parameters of four varieties of lettuces. By building different prediction models, we required predicted value of five phenotypic parameters of lettuce. Test results indicate that prediction models we have constructed are reliable and feasible. In addition,our methods can be better transferred to the research of other crops, and producers can adjust the growing environment of crops in time, so as to obtain higher yield.

The traditional method of measuring the lettuce height is a manual measurement with instruments, which is greatly affected by human error.At present, researchers have proposed to use color cameras to obtain RGB images of lettuce, and to obtain the height of lettuce from the images. However, these tasks usually require camera calibration or a reference object with a known height, which is somewhat restrictive. Considering that deep neural networks have a powerful ability to feature extraction and expression, without camera calibration and reference objects, we try to use four networks of image recognition to explore the effect of deep learning on abstracting the lettuce height from RGB images. On the test set, including 80 images and height from 0.9 cm to 7.5 cm, we achieve a good result with a mean absolute error of 1.22 mm.