Table 3.

Summary of ML methods and corresponding references used in the literature for BW estimation in four livestock species

Reference	ML method	2D/3D Images	Breed	Num. of animals	R	R 2	ARE	MAE	RMSE
Tasdemir et al., 2011a,b	Fuzzy rule-based model	2D	Cattle (Holstein)	115	0.9922
Tasdemir et al., 2019	ANN/MLP	2D	Cattle (Holstein)	115	0.9916
de Moraes Weber et al., 2020	LR SVM regression Regression by discretization with RF	2D	Cattle (Girolando)	34	0.7100			38.4600	46.6900
Gjergji et al., 2020	CNN RNN/CNN RAM RAM with CNN	2D	Cattle (Nellore, Angus)	20				23.1900
Miller et al., 2019	ANN	3D	Cattle (Aberdeen Angus, Limousin, Simmental, Charolais)	1,484		0.7000			42.0000
Cominotte et al., 2020	MLR LASSO PLS ANN	3D	Beef cattle	48		[0.79, 0.92]			[7.78, 18.14]
Fernandes et al., 2020b	MLR PLS ENR MLP DL image encoder model	3D	Pig	557		[0.03, 0.87]		[0.81, 5.20]	[1.05, 6.44]
Rudenko et al., 2020	Mask RCNN network and MLP	3D	Cattle (Ayrshire, Holstein, Jersey, Red Steppe)	500	[0.84, 0.92]		[0.1, 0.17]	[0.9, 2.5]