Research on the Vision-Based Dairy Cow Ear Tag Recognition Method

Area of Science:

• Agricultural Science
• Computer Vision
• Artificial Intelligence

Background:

• Modern dairy farming requires efficient individual cow recognition for precision agriculture.
• Current methods for dairy cow management face challenges with scale and invasiveness.

Purpose of the Study:

• To develop a low-cost, non-invasive, and accurate vision-based automatic recognition system for dairy cow ear tags.
• To enhance precision farming through improved individual animal identification.

Main Methods:

• Proposed a lightweight Small-YOLOV5s model for efficient cow ear tag detection.
• Utilized a Differentiable Binarization network (DBNet) with a Convolutional Recurrent Neural Network (CRNN) for ear tag number recognition.
• Developed and released two standardized datasets for dairy cow ear tag detection and recognition.

Main Results:

• Small-YOLOV5s improved recall by 1.5% and mean average precision by 0.9% over YOLOV5s, with significantly fewer parameters.
• Achieved an impressive 92.1% accuracy in ear tag number recognition.
• Enhanced average prediction speed by 0.5 ms per image.

Conclusions:

• The developed vision-based system offers a cost-effective and accurate solution for dairy cow individual recognition.
• The new datasets will support further research and development in intelligent dairy farming.
• This technology facilitates advancements in the breeding industry through precision management.