Livestock weight measuring and calculating model based on density estimation network

CN122067271APending Publication Date: 2026-05-19珠海市云晓智联科技有限公司 +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
珠海市云晓智联科技有限公司
Filing Date
2026-02-02
Publication Date
2026-05-19

AI Technical Summary

Technical Problem

Existing methods for measuring the weight of poultry and livestock suffer from high stress response, low efficiency, low accuracy, and susceptibility to environmental interference, making it difficult to meet the rapid and accurate measurement needs of large-scale farms.

Method used

A livestock weight estimation model based on a density estimation network is adopted. Through instance segmentation, four-channel image construction, density estimation and weight regression modules, combined with a semantic segmentation backbone network and a multilayer perceptron, a non-contact and highly interpretable weight estimation is achieved.

Benefits of technology

It improves the accuracy and robustness of weight estimation, reduces sensitivity to lighting and background interference, enhances the interpretability and adaptability of the model, and is applicable to different breeding environments and livestock breeds.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122067271A_ABST
    Figure CN122067271A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of livestock and poultry breeding, and particularly relates to a livestock and poultry weight measuring and calculating model based on a density estimation network, which comprises an instance segmentation module, a four-channel image construction module, a density estimation network module, a mass code extraction module and a weight regression module which are connected in sequence. The instance segmentation module is used for detecting and segmenting the input livestock RGB image and outputting a binary instance segmentation mask of the target livestock; the four-channel image construction module is used for splicing the livestock RGB image and the instance segmentation mask in a channel dimension to form a four-channel image tensor as subsequent network input; the density estimation network module is used for receiving the four-channel image tensor; and the weight regression module is used for splicing the n-dimensional quality coding vector and the quantized camera parameters to form a fusion feature vector, and outputting a livestock weight estimation value through a multi-layer perceptron.
Need to check novelty before this filing date? Find Prior Art