What is the role of GEMMO in the FEROX project?

Gemmo serves as a key AI provider in FEROX, primarily focusing on preparing and utilising datasets for berry detection using RGB images. Our responsibility extends to training models specifically for this task. While other partners, notably FBK, also contribute, Gemmo plays a significant role in shaping the business strategy for the potential commercialisation of the project’s outcomes.

How does your expertise contribute to the overarching goals of the FEROX project? 

Leveraging our extensive experience in computer vision, Gemmo brings valuable technical expertise to the FEROX project. As the only startup in the consortium specialising in AI, we also introduce commercial insights. Our role encompasses designing and managing the entire data processing pipeline—from drone-captured data to berry count outputs—addressing both the technological and infrastructural aspects of the project.

How are AI tools evolving based on FEROX experiences?

Our experience in FEROX has streamlined our focus towards object detection algorithms for counting berries, which we found to be simpler and more reliable for our needs than more complex methods like instance segmentation. This refinement in our approach is guiding the development of AI tools, with ongoing discussions about future capabilities, such as running AI models directly on drones, outside the immediate FEROX timeline.