FBK is a private research institution based in Trento (Italy) and operating in different scientific fields and disciplines. As such, it has the role of keeping the Autonomous Province of Trento within the mainstream of international research. FBK is made up of twelve research centers, whose activities and production are available at http://www.fbk.eu.
Workplace
The Digital Industry (DI) Center, one of the Centers of the Bruno Kessler Foundation, focuses its research on digital technologies for the manufacturing industry by creating applications for critical systems, adaptive and autonomous systems, advanced perception, diagnostics, and prediction systems. The research in the DI centre also involves fields related to digital agriculture, cultural heritage, and geomatics. The Technologies of Vision (TeV) research laboratory conducts research in the fields of Computer Vision, Deep Learning, and Robotic Perception. TeV focuses on developing techniques for 2D and 3D scene understanding, including classification, detection, segmentation, tracking, foundation models, large multimodal models, and neural rendering. Its research has applications in various domains:
More information about the TeV lab is available at https://tev.fbk.eu. Project description The internship is carried out within AgRimate, a Horizon Europe project developing AI, augmented reality and robotics solutions for pruning in smallholder olive and vineyard farming. Within the project's vineyard pilot, TeV has already acquired a dataset of grapevines using front- and back-facing cameras and produced independent sparse 3D reconstructions (via COLMAP) for each view. The goal of the internship is to turn before/after image pairs into pseudo-labelled cutting points, to be released as a public benchmark for cutting-point detection/prediction. Core activities include: designing a 2D pipeline to detect differences between before/after image pairs and derive candidate cutting-point pseudo-labels; integrating pseudo-labels across views and projecting them into 3D via multi-view triangulation, with quality control and outlier filtering; establishing baseline methods (heuristic and learning-based) for cutting-point prediction; and packaging the resulting dataset for public release. Extension goals include densifying the 3D reconstructions, merging front/back reconstructions into a single coherent model, and individual plant-instance segmentation. The internship can be fitted into an MSc thesis and offers close mentoring from the TeV group, access to the lab's GPU/cluster compute infrastructure, and the possibility of a technical report or paper draft. Requirements The ideal candidate should have:
Internship Details
Start date: to be agreed, flexible based on the student's academic calendar Duration: from 4 to 6 months Workplace: Trento (Province of Trento, Italy) Benefits: company canteen Application Interested candidates are requested to submit their application by email to Sergio Povoli( spovoli@fbk.eu). The application must include the following documents in .pdf format:
For any further information, please contact: jobs@fbk.eu.