TABLE 1.

An overview of existing image-based methods for lettuce fresh weight monitoring.

Method type	Input data types	Sample sizes	Methods	Descriptions	References
Empirical feature extraction	RGB	/	Traditional image processing + quadratic regression	Regression by projected area from top view images	Lee, 2008
	RGB	82	Traditional image processing + linear regression	Regression by pixel counting from top view images	Jung et al., 2015; Jiang et al., 2018
	RGB	/	OpenCV-based segmentation + linear regression	Regression by extracted 2D and 3D geometric features from a stereo-vision system	Yeh et al., 2014; Chen et al., 2016
	3D point clouds	230	Rule-based segmentation + linear regression	Regression by extracted geometric features from colored 3D point clouds.	Mortensen et al., 2018
	RGB	338	Optical flow analysis + gradient boost regression	Regression by extracted leaf movement features from top view images.	Nagano et al., 2019
	RGB	750	CNN segmentation + linear regression	Regression by extracted geometric features from the side and top view images	Reyes-Yanes et al., 2020
End-to-end deep learning	RGB	286	CNN regression	Regression directly by a CNN model	Zhang et al., 2020
	RGB-D	3,888	CNN regression	Regression directly by an RGB-D fusion CNN network	Buxbaum et al., 2022