Table 9.
An analysis of research conducted for non-contact body measurement applications.
Research | Measurement Method | Device | Object Processing | Animal | Object Measurement | Performance |
---|---|---|---|---|---|---|
Rodriguez Alvarez (2018) [7] | Automatic measurement | Kinect | Depth image | Cow | Body condition score | Accuracy: 78% within 0.25, Accuracy: 94% within 0.5 |
Nir et al. (2018) [2] | Automatic measurement | Kinect | Depth image | Cow | Hip height, withers height | Mean relative absolute error less than 1.17% |
Zhang et al. (2019) [5] | Automatic measurement | Kinect | Depth image | Cow | Measurement points on the backside | Mean absolute error less than 1.17 cm |
Weber et al. (2020) [1] | Automatic measurement | RGB camera | 2D image | Cow | Feature points on the backside | N/A |
Kuzuhara et al. (2015) [24] | Manual measurement | Xtion pro | Point cloud | Cow | Backside | N/A |
Salau et al. (2017) [25] | Manual measurement | Six Kinect | Point cloud | Cow | Teat length, heights of the ischial tuberos | Standard error range are 0.7∼1.5 mm, and 14.0∼22.5 mm |
Le Cozler et al. (2019) [26] | Manual measurement | Five LiDAR sensors | Point cloud | Cow | Volume and surface area | Coefficients of variation were 0.17% and 3.12% |
Song et al. (2019) [27] | Automatic measurement | Three Kinect | Depth image | Cow | Vertebral column, centerline of the sacral ligament, hook bone center | N/A |
Ruchay et al. (2020) [28] | Manual measurement | Three Kinect | Point cloud | Cattle | Withers height, hip height, chest depth, heart girth, ilium width, hip joint width, oblique body length, hip length, chest width | With a 90% confidence level, measurement errors less than 3% |
Our | Automatic measurement | RGB camera | Point cloud | Dairy Cow | Height (stature), rump angle, rump width, front teat length | Height (stature): mean absolute error, 0.7 cm Rump Angle: mean absolute error, 0.61 cm Rump Width: mean absolute error, 2.5 cm Front teat length: mean absolute error, 0.79 cm |