Monocular Underwater Camera Fish Biomass Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating biomass in aquatic livestock are time-intensive, potentially harmful, and do not accurately represent the entire fish population, as they rely on manual removal and weighing of fish, which is inefficient and limited in scope.
Innovation Solution
A method using a single underwater camera to photograph fish, processing images with computer vision and machine learning techniques to estimate biomass based on fish features like truss lengths, eliminating the need for stereo cameras and reducing environmental interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual removal and weighing of fish is used for biomass estimation, then direct measurement of fish weight is achieved, but the process becomes time-intensive and potentially harmful to fish
Solution Approach 1:
The patent replaces the mechanical system of manual fish removal and weighing with an optical-based computer vision system. A monocular underwater camera captures images of fish in their natural environment, and machine learning models automatically estimate biomass from these images, eliminating the need for physical handling and manual weighing processes.
Solution Approach 2:
The patent creates a visual copy of the fish through camera imaging rather than requiring physical interaction. The biomass estimation is derived from image data and learned features (such as truss lengths) that represent the fish's characteristics, allowing indirect measurement without physical contact.
2Ease of operation
If manual sampling of a small portion of fish population is used, then measurement process is simplified, but the true characteristics of the entire population remain unknown
Solution Approach 1:
The patent creates a system that serves multiple functions: it can estimate biomass of individual fish, calculate population-level statistics, identify growth patterns, and detect anomalies. The same computer vision and machine learning infrastructure handles both individual measurement and population analysis, making the system universally applicable to various aquaculture monitoring needs.
Solution Approach 2:
The patent introduces machine learning models and computer vision algorithms as intermediaries between the camera images and biomass estimation. These intermediaries extract meaningful features (truss lengths, body dimensions) from images and translate them into accurate biomass predictions, enabling comprehensive population analysis without direct physical measurement of each fish.
3Measurement precision
If stereo cameras or light-based depth detection hardware is used for depth data, then depth information can be obtained, but hardware complexity and maintenance requirements increase
Solution Approach 1:
The patent extracts the depth perception capability from complex hardware systems (stereo cameras, light-based depth detectors) and implements it through software-based monocular depth estimation. The depth information is derived computationally from single-camera images using machine learning models, removing the need for specialized depth-sensing hardware.
Solution Approach 2:
The patent replaces mechanical/optical depth sensing hardware with a computational approach. Instead of using stereo vision or active light-based depth detection, the system uses a monocular camera combined with machine learning algorithms to estimate depth, substituting physical depth-sensing mechanisms with intelligent image processing.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for monocular underwater camera biomass estimation. In some implementations, an exemplary method includes obtaining an image of a fish captured by a monocular underwater camera; providing the image of the fish to a depth perception model; obtaining output of the depth perception model indicating a depth-enhanced image of the fish data in the image; determining a biomass value estimate of the fish based on the output; and determining an action based on one or more biomass values estimates including the biomass value estimate of the fish.