Neural Radiance Fields (NeRF) are reshaping 3D anomaly detection. By efficiently estimating poses in unseen views and detecting anomalies, NeRFs overcome the computational barriers of previous methods, offering speedier training and inference. This breakthrough is evidenced in the Pose-agnostic Anomaly Detection benchmark and its multi-pose anomaly detection dataset, showcasing state-of-the-art results. Dive into the methods and efficacy of this powerful framework detailed at Arxiv.

• Advances in 3D Gaussian splatting for precise anomaly detection.
• Achieves superior training and inference speed.
• Demonstrates strong performance on the Pose-agnostic Anomaly Detection benchmark.
• Enhances anomaly detection with less training data.