Fresh Scanner

A platform for automating of quality control and categorization of fruits and vegetables with reports.

Our customer is a top manager of one of the largest Russian retailers. He wanted us to create a solution with the following characteristics:

  • Laboratory for digitizing the results of the product examination.
  • Databases with the results of examinations conducted in accordance with the standards of the United Nations Economic Commission for Europe (UNECE) and the Organization for Economic Cooperation and Development (OECD).
  • Artificial intelligence based on trained neural network algorithms for assessing the quality of fresh fruits and vegetables.
  • Software for managing the implementation of the full cycle business process including analysis of products, shipment, and statistics.

Project Facts

6 team members including 3 software developers
80k USD Budget
Revolutionary technology applied


The use of advanced developments in the field of image analysis using:

Convolutional neural networks, industrial applications (RetinaNET, G-NET, MASK-KST)
Canon cameras used in the scanner.
Sets of libraries and frameworks, such as Rutogsh +


  • Accuracy of 96%
  • A scanner
  • Databases conducted with the accordance with the UNECE and OECD standards
  • Product Quality Detection
  • Quality Category Detection
  • Defects Detection
Cortex. Fresh Scanner features



The project is still at the “pre-MVP stage”. These are the achievements which have been already integrated into one platform so far:

  • A scanner for obtaining high-quality photographs has been developed.
  • A unit of trade professionals has been formed to assess the quality of fruits and vegetables.
  • Data collection for machine learning is ongoing.
  • Developed software to assess the quality of fruits and vegetables.
  • The virtual personal Cabinet of the expert-receiver has been developed.
  • The automated process of forming reports has been developed.

Fresh Scanner Development Timeline

The Idea (4Q 2017)

  • We were approached by the client,wanting us to solve his product quality control issue. Once we had a clear understanding of what our client wanted to achieve, an action plan was developed.
Proof of

(3Q 2018- 4Q 2018)

  • After consulting with the industry professionals the technology stack was selected and the timeframe worked out. Project costs were calculated.
  • We developed the brand identity.
  • We registered a company under the name ”Neirofresh”.
  • We supplied the startup company with Programming professionals
  • A prototype of a scanner successfully constructed and tested together with our custom software.

(4Q 2018 - 3Q 2019)

  • During the preparations for the release of the MVP additional tests conducted.
  • The product achieved an accuracy of 96% for the quality category, 43% for defects.
  • MVP soon to be ready.
  • Agreement for further development of the product successfully signed after the latest tests.
  • Company applied for $100k grant.