MinkOcc: Towards real-time label-efficient semantic occupancy prediction

Sze, Samuel; De Martini, Daniele; Kunze, Lars

Abstract

Developing 3D semantic occupancy prediction models often relies on dense 3D annotations for supervised learning, a process that is both labor and resource-intensive, underscoring the need for label-efficient or even label-free approaches. To address this, we introduce MinkOcc, a multimodal 3D semantic occupancy prediction framework for cameras and LiDARs that proposes a two-step semi-supervised training procedure. Here, a small dataset of explicitly 3D annotations warm-starts the training process; then, the supervision is continued by simpler-to-annotate accumulated LiDAR sweeps and images – semantically labelled through vision foundational models. MinkOcc effectively utilizes these sensor-rich supervisory cues and reduces reliance on manual labeling by 90% while maintaining competitive accuracy. In addition, the proposed model incorporates information from LiDAR and camera data through early fusion and leverages sparse convolution networks for real-time prediction. With its efficiency in both supervision and computation, we aim to extend MinkOcc beyond curated datasets, enabling broader real-world deployment of 3D semantic occupancy prediction in autonomous driving.

• Presentation Conference Type: Conference Paper (unpublished)
• Conference Name: 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025)
• Start Date: Oct 19, 2025
• End Date: Oct 25, 2025
• Acceptance Date: Jun 16, 2025
• Deposit Date: Aug 22, 2025
• Peer Reviewed: Peer Reviewed
• Public URL: https://uwe-repository.worktribe.com/output/14832660
• Other Repo URL: https://ora.ox.ac.uk/objects/uuid:729536f5-6be3-4116-9e6a-029e0f970102