What question did this study set out to answer?

This study aims to develop a framework for estimating pig weight using standard RGB images without direct contact.

June 7, 2026Applied SciencesOpen Access

SAM 2-Assisted Vision Transformer and Morphometric Feature Engineering for Pig Weight Estimation from RGB Images

View Full Paper

Ask AI

Bookmark

This study aims to develop a framework for estimating pig weight using standard RGB images without direct contact.
Utilized SAM 2 for foreground extraction combined with a transformer-based dorsal segmentation network.
Introduced cross-covariance attention and local patch interaction modules for segmentation accuracy.
Employed XGBoost regression for weight prediction using 17 extracted morphometric features.
Achieved a mean absolute error of 2.983 kg and an R2 value of 0.9891 on the PIGRGB-Weight dataset.
Reduced prediction error by 24.1% compared to a DeepLabV3+-based baseline.
Demonstrated that enhanced dorsal segmentation improves feature extraction stability from RGB images.

No takes yet. Share an insight, caveat, or question.

Randomized trial demonstrates improved pig weight estimation using RGB images, suggesting better farming practices.

Y et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark