Jaime Corsetti

Highlighted Projects

Functionality Understanding in 3D Scenes

Accepted at CVPR 2025 as Highlight paper

This project focuses on segmenting the interactive components of objects in complex 3D indoor scenes, given a task described in natural language. This represents a fundamental step toward embodied AI agents capable of interacting with their environment. We introduce Fun3DU, the first training-free approach designed for this task, which surpasses the previous state of the art by 11.4 accuracy points.

Paper - Code

Open-vocabulary 6D Object Pose Estimation

Accepted at CVPR 2024 as Highlight paper

Object pose estimation is a fundamental task in computer vision with applications in robotics and augmented reality. We propose a novel approach that uses textual prompts to identify objects, enabling pose estimation for objects never seen during training. Our method, Oryon, achieves state-of-the-art performance on a challenging new benchmark.

Paper - Code

Publications

1. Functionality understanding and segmentation in 3D scenes. Corsetti, J., Giuliari, F., Fasoli, A., Boscaini, D., and Poiesi, F. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025. (website)
2. Open-vocabulary object 6D pose estimation. Corsetti, J., Boscaini, D., Oh, C., Cavallaro, A., and Poiesi, F. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. (website)
3. Revisiting fully convolutional geometric features for object 6D pose estimation. Corsetti, J., Boscaini, D., and Poiesi, F. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, 2023. (website)