Aquaculture 
Phenotyping Machine

This project was developed in partnership with Aquaticode, a company bringing Machine Learning image recognition to the aquaculture industry. The goal for the machine is to generate robust datasets of images to help build the ML models. For this the machine needs to be able to capture several data inputs.

Feature Set Summary
RGB Camera (8MP+)- two perspectives.
Pit-tag scanner (ISO11784 and 11785 134.2 kHz RFID)
Scale(s) to measure 10g to 200g
Barcode Reader
Controlled lighting
Communication with remote servers (cloud or local) over wifi
Components should be able to waistband a wet environment
<5-7 sec processing time per fish
Screen and human interface

Reasearch

The project began with a quick 4 weeks research phase to evaluate a range of technical solutions and approaches. Mainly, to identify:

Controller (Computer)
Camera
PIT Tag Scanner
Lighting
Scale
Structural Design

Development & Testing

After the first successful field test, we were commisioned to manufacture 5 more machines.

So far, the machines have been used in 3 different countries to process thousands of fish.