Entity identification using machine learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
It is challenging to accurately identify and track individual fish, particularly within large populations, due to similarities in appearance and the dynamic nature of fish characteristics over time, which hinders monitoring and management in aquaculture settings.
Innovation Solution
A machine learning model is trained to recognize fish using their unique characteristics, such as spot patterns, and generates embeddings that are mapped to a high-dimensional space for identification and re-identification, allowing for the tracking of fish over time and in real ocean conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional visual inspection methods are used to identify fish, then the process is simple and requires minimal equipment, but the ability to distinguish between individual fish of the same species is insufficient
Solution Approach 1:
The patent replaces manual visual inspection with an automated machine learning-based image recognition system. The system uses trained neural networks to automatically extract and compare fish characteristics from images, substituting human visual capability with computational analysis that can detect subtle patterns invisible to the human eye.
Solution Approach 2:
The system creates digital representations (embeddings) of fish appearances based on image analysis. These embeddings serve as copies or fingerprints of each fish's unique characteristics, allowing for automated comparison and identification without requiring direct visual inspection of the actual fish.
2Duration of action of moving object
If fish characteristics are monitored over prolonged periods, then longitudinal studies and health tracking become possible, but the ability to maintain accurate identification despite changes in fish appearance deteriorates
Solution Approach 1:
The system is designed to dynamically adapt to changes in fish appearance over time. The machine learning model is trained to recognize consistent identifying features (such as spot patterns) while being tolerant of natural variations that occur as fish grow and mature. The embedding comparison approach allows for flexible matching that accounts for developmental changes.
Solution Approach 2:
The system incorporates feedback mechanisms where identification results are continuously refined. By comparing embeddings over time and using training data that includes fish at different life stages, the system learns to maintain reliable identification despite appearance changes, adjusting its recognition criteria based on observed patterns.
3Productivity
If manual tracking methods are used in large-scale fish production, then operational simplicity is maintained, but the ability to track individual fish and their health metrics is insufficient
Solution Approach 1:
The patent replaces manual tracking operations with automated image capture and analysis systems. Cameras positioned in the fish pen automatically capture images, and machine learning algorithms process these images to identify and track individual fish, their growth, and health metrics without human intervention.
Solution Approach 2:
The system performs multiple functions simultaneously: identification, tracking, health monitoring, and data collection. A single automated system handles what would otherwise require multiple separate manual processes, increasing productivity while managing automation complexity through integrated design.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media for identification and re-identification of fish. In some implementations, first media representative of aquatic cargo is received. Second media based on the first media is generated, wherein a resolution of the second media is higher than a resolution of the first media. A cropped representation of the second media is generated. The cropped representation is provided to the machine learning model. In response to providing the cropped representation to the machine learning model, an embedding representing the cropped representation is generated using the machine learning model. The embedding is mapped to a high dimensional space. Data identifying the aquatic cargo is provided to a database, wherein the data identifying the aquatic cargo comprises an identifier of the aquatic cargo, the embedding, and a mapped region of the high dimensional space.