Cengiz Özemli
Akademisyen
- Thread Author
- #1
## High-Speed Imaging System in Poultry Feeding Analysis
Allied Vision's advanced camera system, which provides real-time biomechanical monitoring using AI-based computer vision to analyze the biomechanics of feeding behavior, enables automated data capture in precision livestock farming.
### Precision Livestock Farming with Automation and Artificial Intelligence
In research conducted by Paulista University (São Paulo, Brazil), an AI-powered machine vision system was developed to measure the feeding behavior of broiler chickens. This system contributes to the optimization of feed utilization, which constitutes a significant portion of production costs, by analyzing the biomechanical interaction between animals and feeding systems.
Unlike traditional methods, this technology was designed to provide continuous motion data without using mechanical markers or manual records. This creates a meaningful data flow for digital twin models.
### High-Speed Imaging Setup
The system used in the research includes Allied Vision’s EoSens™ CoaXPress high-speed industrial camera and a Nikon 50 mm f/1.4 lens. The camera was positioned 1.0-1.5 meters from the feeder to record lateral movements at high resolution during feeding.
Operating at 300 fps, the system can detect rapid pecking movements that typical video systems cannot capture. Illumination was standardized between 3000-5000 lux using a 500 W LED light at 6500 K color temperature.
### AI-Based Image Processing
The collected images were processed through a three-stage pipeline running on Python 3.10. YOLOv8 object detection was integrated with the Segment Anything Model (SAM) developed by Meta AI, using PyTorch 2.10.0 and OpenCV 4.13.0.92, to perform anatomical segmentation and motion tracking.
Processing performance was tested on a workstation equipped with an Intel Core Ultra 9 275HX processor, 64 GB DDR5 RAM, and an NVIDIA GeForce RTX 5070 GPU. Sufficient accuracy for automated behavior analysis was achieved with a 95% accuracy rate.
### Relationship Between Feeding Mechanics and Feed Particle Sizes
Data analysis revealed that larger feed particles increased the pecking angle, thereby improving feeding efficiency. This established a directly measurable link between feed particle size and biomechanical effort. This result allows for the optimization of precise feeding strategies.
### Data Foundation for Digital Twins
The system also monitors performance and animal welfare indicators, such as the Beak Efficiency Index, creating a data infrastructure for digital poultry farming models.
The study continues to be developed to enable Edge AI applications in commercial poultry environments, under more complex lighting and obstruction conditions.
---
The content has been rewritten for industry professionals.
