Beef Cattle Body Weight Prediction from Depth Image Biometric Traits

The dataset supports a study predicting body weight and feed efficiency traits in 196 Angus steers using biometric traits extracted from depth camera images. Predictors include projected volume, surface area, length, width, and height, with body weight predicted with high accuracy (R² = 0.96).

Use Cases

• Predict body weight in beef cattle using biometric traits like projected volume, flat surface area, and body length as features.
• Model dry matter intake using image-derived predictors such as projected volume and flat surface area.
• Assess the predictive value of biometric traits like length, width, and height for feed efficiency traits like residual feed intake.
• Compare linear regression performance against complex models like random forest or gradient boosting for traits predicted from biometric data.

Strengths

• Includes data from 196 commercial Angus steers, providing a specific sample size for model development.
• Biometric traits were extracted from depth camera video recordings, offering a non-invasive measurement method.
• Study reports high predictive performance for body weight (R² = 0.96, MAE = 10.12 kg) using specific traits.
• Model performance was evaluated using leave-one-out cross-validation on the available data.

Limitations

• The dataset is small in file size (969.1 KB), indicating limited raw data or a summary document rather than a full tabular dataset.
• Predictive performance was limited for feed efficiency traits like residual feed intake (R² < 0.01).
• The specific column structure and row count of the underlying data are unknown from the provided description.

Provenance

Source

figshare, author Erin Marie Hanson.

Collection Method

Biometric traits (projected volume, surface area, length, width, height) extracted from video recordings of cattle using a RealSense D455 depth camera.

Freshness

Last updated 2026-03-20.