Colony Counter 'IRIS'

Artificial Intelligence Capabilities

In short: if an analyst can recognize a colony in an image within 1.5 seconds, then a neural network can be trained to do the same.
This also applies to distinguishing between particles/debris and colonies — even reflections from colonies can be accurately identified.
A neural network is therefore trained using the knowledge and experience of the analyst — only it's many times faster.

Introduction Program

If you have serious interest, we invite you to be convinced through an introductory program.
At the start of this program, two agar/microorganism combinations are selected in consultation, which will be used to train two neural networks.
A number of representative Petri dishes are then chosen to calibrate the camera and lighting setup. Additionally, we assess whether any existing neural networks are available to assist in the annotation process.

All of these settings are saved as two separate tests in the TrainingSet Creator.
Using this TrainingSet Creator, you will generate images and annotate them.
We provide remote support during the annotation process and monitor the quality of the annotations, as well as the balance and size of the dataset as a whole.
Once complete, we train a new neural network based on your dataset.

This trained neural network will be uploaded for you to evaluate using the Validation Web App.