Live weight prediction of cattle using deep image regression
- Resource Type
- Conference
- Authors
- Ruchay, Alexey; Dorofeev, Konstantin; Kalschikov, Vsevolod; Kolpakov, Vladimir; Dzhulamanov, Kinispay; Guo, Hao
- Source
- 2021 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor) Metrology for Agriculture and Forestry (MetroAgriFor), 2021 IEEE International Workshop on. :32-36 Nov, 2021
- Subject
- Components, Circuits, Devices and Systems
Computing and Processing
Engineering Profession
General Topics for Engineers
Geoscience
Robotics and Control Systems
Weight measurement
Deep learning
Point cloud compression
Machine learning algorithms
Cows
Predictive models
Prediction algorithms
Prediction
live weight
Hereford cattle
deep learning
image regression
- Language
The traditional linear regression algorithm is used to predict the live weight of livestock. However, this traditional method is inadequate for accurate prediction. Recently, a few researchers have successfully applied various machine learning algorithms for predicting the live body weight using livestock morphological measures. We investigate deep learning methods for developing a live weight prediction model based on image regression in this study. We use only RGB images and depth maps for predicting the live cattle weight. The best model for our study is the proposed model with MAPE 9.1% using the RGB images and the depth maps. We have shown results on real-world datasets that demonstrate that the proposed model can reach levels of weight measurement accuracy comparable to those obtained by traditional weighting.