First Open Call: Funded Sub-Projects

The sub-projects funded by the CHAMELEON First Open Call have started.

Five sub-projects will develop new bundles over the course of six months, which will be integrated into the CHAMELEON platform.

These funded sub-projects will be executed by SMEs and Academia with a high capacity for innovation, from various European countries, including Norway, Italy, Portugal, Lithuania, and Spain.

Additional information regarding the funded sub-projects and the objectives of the bundles is available in the following section.

Title: 

Vineyard Leaf Image Analysis for pest and Disease Detection using Explainable Federated Learning.

Company:

NORGES TEKNISK NATURVITENSKAPELIGE UNIVERSITET NTNU

Topic:

Agriculture

Bundle Code:

BC2 - Health status of vineyards and early detection of pest and fungal infestations

Country:

Norway

Abstract:

The VIPA-DELF project proposes a holistic approach to improve pest and disease detection in vineyards using grapevine leaf image analysis through cutting-edge technology and innovative methodologies. The proposed AI approach for early pest and disease detection will leverage the power of SOTA “Computer Vision” based models along with the use of “Federated Learning (FL)” to address the data protection and owner’s privacy concerns, and “Explainable AI (XAI)” to explain the reasoning behind the decisions made Please see Technological Approach” section for the proposed approach.

Final presentation: 

VIPA-DELF

Title:

mandrIAno:AI-empowered stockman 

Company:

ELIF LAB

Topic:

Livestock

Bundle Code:

BC8 - Monitoring livestock/individual animal/virtual fences

Country:

Italy

Abstract:

The mandrIAno (Artificial Intelligence-empowered stockman) project will adopt a semantic computer vision system that, starting from the videos collected using a drone, is able to produce a report in natural language in which the artificial intelligent component provides information related to the positioning of the monitored animals (in association with data provided by GPS); the count of animals in a group; and describes, through the use of a transformer-based visual question answering model and a language model component, the environment in which the animals are detected, highlighting the context and environmental features of the scene.

Final presentation: 

mandrIAno

Title:

Sustainable Aerial Forestry Resilence Analytics

Company: 

FORCERA, LDA

Topic:

Forestry

Bundle Code:

BC4- Early detection of health status in forest

Country:

Portugal

Abstract:

SAFRA is focused on developing a comprehensive software bundle that simplifies the collection of analytical data from the analysis of RGB images collected by drones in forest environments. SAFRA’s primary objective is to create a user-friendly platform that facilitates collecting and analysing analytical insights from these images, with a particular emphasis on assessing forest health. The SAFRA bundle will combine sensorial fusion from Unmanned Aerial Vehicles (UAVs) and Remotely Piloted Aircraft Systems (RPAs), open data, AI, and Digital Twin technology to deliver a comprehensive yet intuitive analysis of the health status of forests, combining the expertise of the proponent company, FORCERA, with state-of-the-art open-source tools and frameworks. The software bundle's key features include the integration of geolocation and meteorological data, along with the utilisation of this information to adjust RGB image contrast and saturation collected by UAV/RPA devices. By combining imaging with precise location data and meteorological conditions, the system aims to enhance the precision of the detection of indicators of poor leaf health in trees and vegetation and assess the impact of factors such as drought and pests on the ecosystem.

Final presentation: 

SAFRA

Title:

THRUST LOG-IQ: Image-based Quantification of Logs

Company:

UAB AeroDiagnostika (THRUST - Intelligent UAV Systems)

Topic: 

Forestry

Bundle Code: 

BC1 - Access to forest (paths, roads)

Country: 

Lithuania

Abstract: 

THRUST LOG-IQ vision is to develop a software that would be able to identify large trees fallen on forest roads in order to evaluate forest accessibility after extreme weather events (BC1) and then expand this functionality for the software to be able to assess wind damage in whole forest area and account as well as classify deadwood for biodiversity conservation.

Final presentation: 

THRUST LOG-IQ

Title: 

Timber stack Inventory for Logging Operations with UAVs

Company: 

PANOimagen

Topic:

Forestry

Bundle Code: 

N/A

Country:

Spain

Abstract:

Project TILO aims to address the remote monitoring of Logging Operations in isolated and rural areas, by assessing the advancement of the exploitation on a regular basis via UAV flights. TILO will quantify the extension that has already been harvested and the wood volume in timber stacks waiting to be transported.

Final presentation: 

TILO