Related Experiment Video
Updated: Nov 19, 2025
08:32
Tracking Rats in Operant Conditioning Chambers Using a Versatile Homemade Video Camera and DeepLabCut
Published on: June 15, 2020
12.9K
Now You See Me: Convolutional Neural Network Based Tracker for Dairy Cows.
Oleksiy Guzhva1, Håkan Ardö2, Mikael Nilsson2
1Department of Biosystems and Technology, Swedish University of Agricultural Sciences, Alnarp, Sweden.
Frontiers in Robotics and AI
|January 27, 2021
Summary
Researchers developed a new method using convolutional neural networks (CNNs) for tracking individual dairy cows. This automated system accurately identifies and monitors cows, crucial for improving herd health and welfare.
Area of Science:
- Animal Science
- Computer Vision
- Agricultural Engineering
Background:
- Maintaining dairy cattle health and welfare necessitates analyzing cow-to-cow behaviors.
- Accurate individual tracking is essential for on-site behavioral analysis in dairy farms.
Purpose of the Study:
- To introduce a novel method for continuous tracking and identification of individual dairy cows.
- To leverage convolutional neural networks (CNNs) for robust animal monitoring.
Main Methods:
- Utilized top-down view recordings from three ceiling-mounted cameras over four months (500 million frames).
- Implemented a CNN-based system to track and identify 252 Swedish Holstein cows in a 6x18 meter waiting area.
- Evaluated system accuracy by comparing tracker outputs with gate data for 26 individual cow tracks.
Main Results:
- Achieved 23 out of 26 tracks correctly identified.
- The system maintained correct cow positioning for an average of 225 seconds per individual.
- Successful tracking exceeded 20 minutes in mildly crowded scenes (<10 cows).
Conclusions:
- The proposed CNN-based system is a significant advancement for automated dairy cow monitoring.
- This technology enables continuous observation of cow behavior and interactions.
- It serves as a foundation for improving dairy farm management and animal welfare.
