Natural objects rendering for economic AI models

Deadline

Natural objects rendering for economic AI models

Challenge

Natural objects (vegetables, fruits, food, etc.) are omnipresent in different industrial applications: food sorting, vegetable treatments, surroundings perception, recognition, etc. These applications require the use of Vision and AI models to get the tasks done reliably and robustly. However training these AI models would require thousands of images / videos with detailed annotations of different items. In the state of the arts, one needs > 10kimages to train an AI model with an accuracy of > 90%, when, in average one minute is needed to annotate one image. The more the variability one wants to cover, the more training images are needed. With successful results from PILS SBO, where rendering techniques were applied to industrial products with CAD information, to retrieve AI training data from updated CAD models, the goal of this project is to extend this research to Natural (no CAD available) objects  rendering techniques to generate large & variable data for economic AI models training and for a reliable & robust outputs. The industrial applications cover amongst others: fruits & vegetables processing & sorting, advanced monitoring of crop & vegetables (such as spray treatment of good crop & discrimination from the bad one), precision farming (such as automatic picking of mature fruits), etc. The applications cover both indoor and outdoor conditions with their variabilities. 

Project goals

NORM.AI aims to develop an innovative solution combining photorealistic virtual 3D environments to automatically generate labeled training data and beyond state-of-the-art artificial intelligence, fed and validated by real-world domain knowledge in industrial applications such as, fruits & vegetables processing & sorting, advanced monitoring of crop & vegetables (e.g., spray treatment of good crop & discrimination from the bad one), precision farming (e.g., automatic picking of mature fruits), …

The project's research will help to deliver the enabling technology for the Flemish technology providers to foster and promote new R&D on agricultural robotics and orchard inspection systems in order to achieve more efficient long-term resource-efficient sustainability of agro-food production in Flanders and reduce the barriers that prevent widespread adoption of robots.

NORM.AI proposes to remove the bottlenecks with an innovative solution combining photorealistic virtual 3D environments to automatically generate labeled training data and beyond state-of-the-art artificial intelligence, fed and validated by real-world domain knowledge. nesy

The project's research will help to deliver the enabling technology for the Flemish technology providers to foster and promote new R&D on agricultural robotics and orchard inspection systems in order to achieve more efficient long-term resource-efficient sustainability of agro-food production in Flanders and reduce the barriers that prevent widespread adoption of robots.

Project goals

NORM.AI_SBO is a Strategic Basic Research (SBO) project. We are looking for companies to join the User Group and work with us on the valorisation of the project.

Interested? Complete the form below and we will contact you as soon as possible.

See our other open calls

Natural objects rendering for economic AI models

Deadline

Strategic Basic Research (SBO)
Robust realization of fractional order control for uncertain general events

Deadline

Strategic Basic Research (SBO)
Efficient programming of robots via learning in a VR environment

Deadline

Strategic Basic Research (SBO)