A groundbreaking dataset has been unveiled, promising a transformative shift in the working conditions of wild berry pickers in Finland.

Finland, renowned for its wild berries, witnesses a natural growth of approximately 500 million kg of these fruits annually. However, a mere 10% of this bounty is harvested, primarily due to the challenges posed by manual forest harvesting. The short berry season often sees foreign workers, unfamiliar with the Finnish language, culture, and terrain, taking on the labour-intensive task.

To address these challenges, the European Commission-funded FEROX project was initiated. Spearheading AI development, Dublin-based AI firm, Gemmo, alongside eight other tech and academic partners have come together to develop a new way of monitoring workers. Their mission is to harness the power of autonomous drones and AI to enhance the working conditions of these pickers.

Over a span of three days in Illomatsi, Finland, diverse data collection methods were employed. From standard RGB images to multispectral shots, and even high-resolution videos, the team left no stone unturned. Drones, flying approximately 2 meters above ground, were pivotal in capturing these images and mapping the forest terrain. Additionally, pickers were equipped with body cameras, and IoT devices like heart rate monitors, GNSS trackers, and eye-tracking devices were introduced, broadening the data acquisition spectrum.

Comprising 1,000 berry images and extensive area mapping visuals, the dataset's primary goal is to replicate real-world conditions. The data will be used to train AI models to identify and count berries, enabling the creation of heat maps to guide pickers to berry-rich zones. This not only aids in efficient harvesting but also minimises ecological disturbances.

Gemmo's meticulous annotation process ensures the dataset's quality. From manual image annotation to geotagging using drone GPS data, every step is designed to provide precise spatial information. The ultimate aim is to differentiate various berry species and generate species-specific heat maps, offering pickers a clear visual representation of berry distribution.

The dataset, set to be published for AI utilisation, holds immense potential for similar berry projects worldwide. By leveraging this data, the FEROX project aims to create detailed spatial distribution maps of berry species and train deep learning models for automated heat map generation. This will empower berry pickers to strategise their harvesting, ensuring maximum yield with minimum environmental impact.

Gemmo AI
77 Sir John Rogerson's Quay - Block C Dublin Docklands, Dublin 2, Ireland
Johanna@gemmo.ai

Founded by Dr. Luca Marchesotti in 2014, we are headquartered in Dublin, Ireland with an office in Milan, Italy, and London, UK.

Since then, Our leadership team has shepherded countless R&D projects across the finish line for Fortune 500 companies - yielding groundbreaking products and services that have transformed entire industries.

Through this work, we have become experts in delivering AI solutions for the manufacturing, environmental monitoring and sports tech sectors.

This release was published on openPR.