3D Animal Behavior Classification Using Unsupervised Depth Imaging

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for animal behavior analysis are subjective, biased, and limited by the use of 2D cameras, which are arena-specific and lack comprehensive assessment of olfactory function, particularly in the context of Autism Spectrum Disorders (ASDs) and neurodegenerative disorders, requiring human intervention and curated databases.

Innovation Solution

A system using affordable 3D depth cameras and custom software to classify animal behavior objectively, segmenting animals from backgrounds, determining orientation, and quantitatively describing their three-dimensional contours, location, and morphological descriptors, employing clustering algorithms to identify Quantitative Behavioral Primitives (QBPs) without human supervision.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If manual observation and hand-annotated databases are used for behavior classification, then subjective evaluation allows flexible anthropomorphic categorization, but reliability and reproducibility decrease due to human bias and limited throughput

Engineering Contradiction:
Improveflexibility in behavior categorizationVSAvoidreliability and reproducibility of results
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The system performs self-service by automatically discovering and characterizing animal behaviors through unsupervised clustering algorithms without requiring human observers to pre-define behavior categories. The computer system independently identifies behavioral patterns from raw video data, eliminating human bias while maintaining adaptability to diverse behaviors through data-driven classification

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical system of manual observation with an automated computerized vision system that processes video data through algorithms. This substitution eliminates human limitations in throughput and consistency while maintaining the ability to categorize behaviors through objective computational analysis rather than subjective human judgment

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Device complexity

If 2D arena-specific cameras are used for video acquisition, then system complexity is reduced, but throughput is limited and alignment errors increase

Engineering Contradiction:
Improvesimplicity of video acquisition systemVSAvoidthroughput of behavioral analysis
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The patent implements universality by using a standardized 2D camera system that can acquire video data across multiple behavioral arenas and experimental conditions without requiring arena-specific configurations. This multi-functional approach increases throughput by eliminating the need for separate specialized systems while maintaining adequate measurement capability through consistent coordinate transformation algorithms

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Device complexity

If human-curated parametric databases are used for behavior matching, then assessment is limited to identifiable behaviors, but system complexity and setup time are reduced

Engineering Contradiction:
Improvesimplicity of analysis systemVSAvoidbehaviors that defy simple categorization
Core Design Contradiction:
Device complexityVSLoss of information

Solution Approach 1:

The system performs preliminary action by automatically generating behavioral classifications from raw video data before any human interpretation is needed. The unsupervised clustering algorithms pre-process the data to identify patterns, which can then be validated and labeled by researchers, reversing the traditional workflow where human observers first define categories and then match behaviors to them

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent adds another dimension to behavior analysis by transitioning from discrete pre-defined categories to continuous multidimensional behavioral space through clustering algorithms. This dimensional expansion allows the system to capture subtle behavioral variations and novel patterns that defy simple categorization, representing behaviors as points in a continuous parameter space rather than forcing them into fixed bins

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentEP4198926B1Method and apparatus for automatically discovering, characterizing, classifying and semi-automatically labeling animal behavior and quantitative phenotyping of behaviors in animals
Publication Date: 2025.11.05 PRESIDENT & FELLOWS OF HARVARD COLLEGE
  • EP4198926B1 patent drawingFigure 1A~1H
  • EP4198926B1 patent drawingFigure 2A~2B
  • EP4198926B1 patent drawingFigure 3

AI summary

A method is provided for studying the behaviour of an animal in an experimental area using a camera to obtain a video stream having a plurality of images of the area, removing background noise from the images, extracting parameters from the images to form multidimensional data points, clustering data points so that each cluster represents and animal behaviour and outputting a matrix of results. Apparatus for performing the method is also provided.