Other researchers previously investigated different approaches
to estimate weight of pigs using image analysis.
Brandl and Jorgensen (1996) used spline functions to express the relationship
between the body area of the pig measured by image analysis and
the live weight of the pig. Marchant et al. (1999) developed automated
algorithms that could find the plan view outline of pigs in
a normal housing situation, measure major body components
and predict the weight of the group of pigs at 34 kg with standard
errors of 7.3% while using manual weighing to calibrate the system.
Schofield et al. (1999) developed prototype imaging systems
to record the weight-related areas of pigs by fitting linear regression
coefficients. Furthermore, they could log the growth rates of
three groups of pigs of three genetic strains to within 5%.
Whittemore and Schofield (2000) examined the value of the estimation
of size and shape for animal description in relation to nutrient
use in breeding sows and growing pigs. Craig and Schinkel
(2001) proposed a mixed effects model1 to estimate pig weight.
White et al. (2004) used a VIA system to continuously collect size
and shape data of a total of 116 pigs from 25 to 115 kg of weight
for three types of pigs and could classify these groups in 64–83%
of observations.