CYOCEAN
AQUAPARK™ / UNDERWATER VISION & AI

See Clearly Below,

Read Growth Better.

Underwater Vision & AI
From image enhancement in complex water to fish measurement and school observation, making underwater perception a grounded basis for farming decisions.
View Core CapabilitiesConsult the Tech Team
Image Enhancement / 3D Measurement / School Insight
FIVE MODULES. ONE CONNECTED VIEW.

From underwater perception to actionable farming insight.

Starting from images, we link fish-body scale, school state, and operational value into a continuous information chain — from sensing to decisions.

01

Image Enhancement

Noise removal · detail recovery

02

Stereo Measurement

Individual metrics · school density

03

Density Quantification

Spatial distribution · school changes

04

Behavior Recognition

School state · anomaly detection

05

Economic Assessment

Cost accounting · ops guidance

01 / IMAGE ENHANCEMENT

Let the details hidden by water re-emerge.

Targeting light attenuation, water scattering, and colour shift, a five-stage preprocessing pipeline improves imaging, providing clearer input images for downstream fish recognition.

01
02
03
04
05
Five-stage preprocessing pipeline
  • 01
    Bilateral Filter Denoising
  • 02
    Gamma + Grayscale Stretching
  • 03
    Dark-channel De-scattering / Dehazing
  • 04
    CLAHE Local Contrast Boost
  • 05
    USM Lightweight Detail Sharpening

Underwater vision input sample: processing effects need to be verified against water conditions.

Clear imaging is not the destination — it is the starting point that makes recognition and measurement more reliable.

02 / STEREO VISION & BIOMASS

From 2D images to real scale.

Solving 3D coordinates via stereo disparity and correcting perspective distortion, turning keypoints into body length and per-fish biomass data, providing a basis for growth tracking and sonar model calibration.

Aquapark™ / Underwater School Sample
Material Sample

Underwater fish keypoint detection and length / mass estimation · technical material frame

  • 01
    Keypoint Detection
    Identify fish keypoints, building target features needed for measurement.
  • 02
    Stereo 3D Reconstruction
    Recover spatial scale via stereo camera disparity, estimating true fish body length.
  • 03
    Per-fish Biomass Estimation
    Combined with a random forest regression model, calibrating the sonar prediction model.
Vision Measurement

Body length and per-fish biomass estimation provides ground truth reference for acoustic models.

Acoustic Supplement

When vision is limited by turbidity or low light, multi-beam sonar supplements dim scanning.

03—04 / SCHOOL INTELLIGENCE

Not just seeing schools, but seeing how their state changes.

Visualizing spatial distribution, swimming speed and group characteristics, observing density changes over time and surfacing stress-escape, over-aggregation, and stagnation anomalies.

Density changes, trace by trace

Count fish projection ratio per unit water, presenting school distribution and changes via spatial density heatmap.

Turning anomalies into visible signals

Extract swimming speed, orientation, and group characteristics, turning abnormal behavior into actionable farming alerts.

Heatmaps and trajectories are interface illustrations — not on-site measurement results.

05 / ECONOMIC INTELLIGENCE

Connect every bit of growth to operational value.

Linking biomass, survival rate, body length and market price, placing batch cost, output and operational trends in one view, supporting feeding, harvesting, and sales planning.

  • 01
    Monitoring Inputs
    Biomass · survival · body length
  • 02
    Cost & Price
    Farming cost · market price
  • 03
    Operational Outputs
    Total output · net profit · profit margin · ROI
  • 04
    Operations Advice
    Feeding · harvesting · sales
Curves are for illustration of information hierarchy and UI form only — they do not represent actual economic calculation.
BUILT ON OBSERVATION. VALIDATED WITH DATA.

Engineering judgment with process and evidence.

From sample conditions to control measurements, jointly defining the scope of technical validation. Conversations can start with the materials below.

Record species and body specs
Document water, lighting, and imaging conditions
Sample conditions
Species · specs · water
Define sample labeling rules
Prepare manual measurement and control data
Measurement basis
Labeling · measurement · control
Cross-check model output vs. measurement
Record error, conditions, and applicable scope
Validation log
Process · result · scope

Validation scope and effects should be confirmed against sample and on-site conditions. The tech team is available to discuss required materials.

FROM UNDERWATER VISION TO BETTER DECISIONS

Let underwater change become the basis for the next step.

From underwater imaging to school insight, let's discuss the farming scenario and technical path that fit you.

Talk to the Tech Team