Livestock Counting via AI Segmentation Masks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for counting livestock, such as physical tagging and manual counting, are invasive, stressful for animals, and prone to errors due to tag damage or loss, and lack accuracy in large-scale herding.
Innovation Solution
A system utilizing image capture devices and artificial intelligence (AI) to process video feeds of livestock walking along a path, employing convolutional neural networks and region proposal networks to identify and count individual animals based on feature maps and segmentation masks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physical tagging systems are used to identify and count livestock, then individual animals can be tracked, but the system causes stress to animals and tags can be damaged or lost reducing reliability
Solution Approach 1:
The patent replaces mechanical/physical tagging systems with an optical/image-based detection system. Instead of using physical tags that cause stress and can be lost, the system uses image capture devices to detect and identify livestock through visual recognition, eliminating the need for intrusive physical attachments while maintaining reliable identification.
Solution Approach 2:
The patent creates a digital copy or representation of livestock appearance through image capture. By capturing images and extracting visual features, the system creates a digital profile of each animal that can be used for identification and counting, replacing the need for physical tags with an optical copy of the animal's visual characteristics.
2Object-affected harmful factors
If manual counting methods are used on samples of livestock, then counting can be performed without stress to animals, but accuracy is reduced due to extrapolation from small samples
Solution Approach 1:
The patent changes the parameter of observation from partial/sample-based counting to comprehensive/complete counting. By using image capture devices to detect all livestock within the field of view and applying computer vision algorithms to identify and count each individual animal, the system transforms the counting process from sampling to complete enumeration, thereby improving measurement precision without increasing stress to animals.
3Device complexity
If traditional counting systems are used, then implementation is simple, but real-time counting capability and accuracy in large-scale herding are insufficient
Solution Approach 1:
The patent replaces manual counting operations with automated computer vision processing. Image capture devices continuously capture footage of livestock, and AI algorithms automatically process these images to identify, track, and count individual animals in real-time. This substitution of mechanical/manual processes with automated optical and computational systems enables real-time counting capability while managing the complexity through integrated software-hardware solutions.
Data Source
AI summary
A system configured to receive video and/or images from an image capture device over a livestock path, generate feature maps from an image of the video by applying at least a first convolutional neural network, slide a window across the feature maps to obtain a plurality of anchor shapes, determine if each anchor shape contains an object to generate a plurality of regions of interest, each of the plurality of regions of interest being a non-rectangular, polygonal shape, extract feature maps from each region of interest, classify objects in each region of interest, in parallel with classification, predict segmentation masks on at least a subset of the regions of interest in a pixel-to-pixel manner, identify individual animals within the objects based on classifications and the segmentation masks, and count individual animals based on identification, and provide the count to a digital device for display, processing, and/or reporting.