Get Started
Home
Topics
Search
Library
Research questionHow can semantic occupancy systems detect unknown objects under distribution shift in autonomous-driving scenes?Semantic occupancy systems represent driving environments densely, but objects unlike those seen during training may be misclassified or missed. Detecting such objects requires identifying anomalies without discarding the surrounding occupancy interpretation.
AI
Computer Vision
Evaluation & Benchmarks
Machine Learning
Robotics
Latest papersRecent research connected to this question, newest first.Out-of-Distribution Semantic Occupancy PredictionThe source integrates unknown-object detection into 3D voxel-based semantic occupancy prediction and uses complementary voxel and bird’s-eye-view representations. Evidence comes from VAA-KITTI and VAA-KITTI-360 with synthetically injected anomalies, so generalization beyond these datasets and settings is not established.research paper · Sep 4, 2026
Related questions
How can one 3D semantic occupancy model handle indoor and outdoor scenes with varying cameras, scales, and label taxonomies?How can 3D occupancy models learn from noisy 2D pseudo-labels without 3D annotations?How can we predict local 3D occupancy from a single underwater RGB image despite unreliable visual geometry?How can we semantically compare autonomous-driving image subsets at scale and attribute their differences to objects?