E3 Project

MICOVISION

Brands
  • Title of the main project: FOODECO: Strategic alliance of the agri-food sector for the complete renewal of its value chains, through its ecological and digital transformation by means of innovative solutions
  • File Number: PAG-010000-2023-18
  • Title of the primery project: MICOVISION
  • Program: Actions to strengthen the industry in the agri-food sector within the Strategic Project for the Economic Recovery and Transformation of the Agri-Food Sector (PERTE Agri-Food Sector) within the framework of the Recovery, Transformation and Resilience Plan (PRTR) in 2023 (Order ICT/738/2022, of July 28, establishing the regulatory bases for granting aid for actions to strengthen the industry in the agri-food sector within the Strategic Project for the Economic Recovery and Transformation of the Agri-Food Sector, within the framework of the Recovery, Transformation and Resilience Plan, modified) by Order ICT/1307/2022, of December 22 and Order ICT/148/2023, of February 16).
  • Project Location: Paterna, Valencia (España)
  • Start Date: 01/03/2023 – End Date: 30/06/2025
  • Partially funded by: MINISTERIO DE INDUSTRIA, COMERCIO Y TURISMO – DIRECCIÓN GENERAL DE ESTRATEGIA INDUSTRIAL Y DE LA PEQUEÑA Y MEDIANA EMPRESA
  • Name of the interlocutor in the Administration: ADDED VALUE SOLUTIONS S.L.

Origin

MICOVISION is one of the primary projects (PP26), belonging to the FOODECO Tractor Project within the AGRO-ALIMENTARY PERTE – Fruit and Vegetable Value Chain. FOODECO’s main objective is to generate a renewal of the ecosystem of the different value chains of the agri-food industry to carry out a digital and ecological transformation through R&D&I activities in the different value chains: Fruit and Vegetable Value Chain; Fishing Value Chain; Olives Value Chain; Meat Products Value Chain; Transversal Machinery Value Chain; Transversal Technologies Value Chain.

MICOVISION Objectives

MICOVISION is an applied research project whose main objective is the applied research of an autonomous system for analyzing the risk of fungal contamination in aromatic plants intended for the production of herbal teas and seasonings, through the application of artificial vision systems with artificial intelligence in the form of Multi-Axis Vision Transformers (MaxViT) that allows improving food safety and reducing the proliferation of biological elements that can produce mycotoxins harmful to the human and animal organism. With this, an Industrial Research project will be carried out focused on the investigation of disruptive technologies in the form of Artificial Vision and Artificial Intelligence systems such as Multi-Axis Vision Transformers for the improvement of food safety derived from the proliferation of unwanted biological elements in the form of mycotoxins.

Objectives related to the main objective of the project:

Throughout the execution of the project, different investigations related to the objectives of the project will be carried out:

  • Research into different Artificial Vision (VA) technologies likely to provide high reliability in the detection of fungi and their proliferation cycles, analyzing the applicability of photonic techniques for image capture, including the use of hyperspectral images, infrared thermography or fluorescence.
  • Applied research on different architectures, models, techniques and AI learning algorithms, to determine those that allow the identification of contaminating fungi that can produce mycotoxins from digital images of plant material and decision making based on that information (transfer learning, data augmentation, dropout, regularization, etc.).
  • Research, design, training and evaluation of Deep Learning models, from different types of potentially available images based on innovative architectures of the “Multi-Axis Vision Transformers” type and the possible hybridization of different architectures.
  • Research on Artificial Intelligence technologies of “Multi-Axis Vision Transformers” type architectures. • Research into Artificial Intelligence technologies for hybridization of “Vision Transformers” (ViT)”, “Convolutional Neural Networks” (CNN) and “Multi-Axis Vision Transformers (MaXViT)” architectures
  • Design and conception of an AI tool that allows the implementation of learning models of novel technologies based on Multi-Axis Vision Transformers and multi-layer Deep Learning for the identification of contaminating organisms in the form of mycotoxin-producing fungi from digital images of plant raw material and taking measures based on that information
  • Validation of the best model obtained in the form of a proof of concept to test its potential usefulness in identifying the fungus Aspergillus flavus.
  • Analysis of the model’s performance, evaluation of results and generated knowledge.

 Participants

MICOVISION involves companies from both the agri-food sector: GENERANDI, S.L. and GENERANDI AGRO, S.L., and the information technology and computational science sector: INVESTIGACIÓN Y DESARROLLO INFORMÁTICO EIKON, S.L.U. and KENUS INFORMÁTICA, S.L.