Bridging 3D Anomaly Localization and Repair via High-Quality Continuous Geometric Representation

Conference: ICCV 2025 arXiv: 2505.24431 Code: https://github.com/ZZZBBBZZZ/PASDF Area: 3D Vision / Anomaly Detection / Point Cloud Keywords: 3D anomaly detection, signed distance function, pose alignment, anomaly repair, point cloud

TL;DR

This paper proposes the PASDF framework, which achieves continuous geometric representation via a pose-aware signed distance function (SDF), unifying 3D anomaly detection and repair tasks, and attains state-of-the-art performance on Real3D-AD and Anomaly-ShapeNet.

Background & Motivation

Importance of 3D anomaly detection: In manufacturing quality control, robotic manipulation, and related fields, even minor 3D anomalies (missing features, deformations, geometric irregularities) can cause entire component failures, necessitating robust 3D anomaly detection techniques.

Limitations of Prior Work: - Voxel-based methods: Discretization causes loss of fine geometric detail and suffers from cubic memory growth - Point cloud methods: Sparse sampling leads to density inconsistency and incomplete surface coverage - Projection-based methods: Information loss in occluded regions and viewpoint-dependent distortions - All of these methods are fundamentally discrete representations that introduce quantization artifacts, which are detrimental to fine-grained anomaly localization

From detection to repair: In 3D printing and advanced manufacturing, anomaly detection is only the first step; in-situ repair is equally important. Conventional methods rely on indirect feature mappings and cannot provide explicit shape reconstruction to guide repair. Even reconstruction-based approaches (e.g., IMRNet, R3D-AD) fail to generate continuous, high-fidelity repair templates due to their reliance on discrete point cloud representations.

Core Motivation: To leverage the continuity and smoothness of SDFs to bridge anomaly detection and repair, while decoupling pose to address detection under arbitrary orientations.

Method

Overall Architecture

PASDF consists of three core stages:

1. Pose Alignment Module (PAM): Aligns input point clouds to a canonical coordinate system, eliminating the influence of pose variation
2. SDF Network: Learns a continuous signed distance function representation that implicitly captures object geometry
3. Anomaly Scoring Module: Computes anomaly scores based on the deviation of SDF values for test samples

Theoretically, anomaly detection is formulated as evaluating the likelihood that a test sample conforms to the distribution of normal shapes. Pose invariance is achieved by integrating over the SE(3) group, approximated by alignment to a canonical pose.

Key Designs

1. Pose Alignment Module (PAM)

PAM adopts a coarse-to-fine two-stage registration strategy:

• Coarse alignment: Applies voxel downsampling to the raw point cloud, extracts FPFH (Fast Point Feature Histogram) features, and performs global coarse registration via RANSAC
• Fine alignment: Refines the coarse registration result using the ICP (Iterative Closest Point) algorithm
• Iterative refinement: Introduces a Chamfer Distance-driven feedback mechanism that dynamically adjusts the loss threshold \(\tau\) to avoid local minima. The cumulative transformation matrix is updated iteratively: \(T^{(k)} = T_{icp}^{(k)} \cdot T_{ransac}^{(k)} \cdot T^{(k-1)}\)

Key PAM parameters: loss threshold \(\tau = 0.016\), increment \(\Delta\tau = 0.001\), maximum iterations \(K = 10\).

2. SDF Network

The aligned point cloud is parameterized via a neural network as an SDF representation:

• Query points are sampled as surface points (10k), points inside the bounding box (10k), and unit volume points (3k), totaling 23,000 points
• Sinusoidal positional encoding \(\gamma(\mathbf{x}_i)\) is applied to query point coordinates
• Network architecture: 8-layer MLP with weight normalization, ReLU activations and 0.2 dropout in intermediate layers, skip connection at the fourth layer
• SDF predictions are trained with a truncated L1 loss

3. Anomaly Score Computation

• Point-level anomaly score: \(A(\mathbf{x}_j) = |f_\theta(\mathbf{x}_j)|\), i.e., the absolute value of the SDF
• Object-level anomaly score: Mean of the top-\(K\) (\(K = 1000\)) highest point-level anomaly scores

4. Anomaly Repair

The trained SDF network implicitly encodes a "normal" shape manifold, which is exploited as follows: 1. The anomalous input is aligned to the canonical pose via PAM 2. The Marching Cubes algorithm extracts the zero-level isosurface from the SDF 3. A point cloud is sampled from the resulting triangular mesh as the repaired output

Loss & Training

• Truncated L1 loss: \(\mathcal{L}_{SDF} = \frac{1}{N_q} \sum_{i=1}^{N_q} |\text{clamp}(\hat{s}_i, -d_{max}, d_{max}) - s_i|\)
• Truncation distance \(d_{max} = 0.1\)
• Training for 2000 epochs with learning rate \(1\times10^{-5}\)
• Data preprocessing: coordinates normalized to \([0,1]^3\); non-manifold detection and Poisson reconstruction applied to ensure watertightness

Key Experimental Results

Main Results

Datasets: - Real3D-AD: A high-resolution real-world dataset with 12 categories, 4 normal training samples and 100 test samples per category - Anomaly-ShapeNet: A synthetic dataset with 40 categories and over 1,600 samples

Results on Real3D-AD

Method	O-AUROC ↑	P-AUROC ↑
BTF(Raw)	0.635	0.571
BTF(FPFH)	0.603	0.733
M3DM(PointMAE)	0.552	0.637
PatchCore(FPFH+Raw)	0.682	0.680
RegAD	0.704	0.705
IMRNet	0.725	-
Group3AD	0.751	-
PASDF (Ours)	0.802	0.745

PASDF surpasses Group3AD by 5.1% in O-AUROC, with particularly strong performance on Seahorse (1.000), Car (0.959), and Fish (0.989).

Results on Anomaly-ShapeNet

Method	O-AUROC Mean ↑	P-AUROC Mean ↑
BTF(Raw)	0.493	0.550
M3DM	0.552	0.616
CPMF	0.559	-
RegAD	0.572	0.668
IMRNet	0.661	0.650
R3D-AD	0.749	-
PASDF (Ours)	0.900	0.897

PASDF achieves the best O-AUROC in 37 out of 40 categories, with perfect scores of 1.000 in multiple categories.

Ablation Study

Effect of PAM on Different Baselines (Anomaly-ShapeNet)

Method	PAM	O-AUROC ↑	P-AUROC ↑
BTF(FPFH)	✗	0.528	0.628
BTF(FPFH)	✓	0.579	0.683
PatchCore(FPFH)	✗	0.568	0.580
PatchCore(FPFH)	✓	0.814	0.867
PatchCore(PointMAE)	✗	0.562	0.577
PatchCore(PointMAE)	✓	0.626	0.681
PASDF (Full)	✓	0.900	0.897

Incorporating PAM into PatchCore(FPFH) yields a remarkable gain of 24.6% in O-AUROC and 28.7% in P-AUROC.

Component Ablation (Anomaly-ShapeNet)

Method	O-AUROC ↑	P-AUROC ↑
w/o RANSAC	0.711	0.739
w/o ICP	0.727	0.836
w/o iterative refinement	0.871	0.884
w/o positional encoding	0.887	0.783
PASDF (Full)	0.900	0.897

Anomaly Repair Quality Evaluation

Method	Real3D-AD CD ↓	Real3D-AD EMD ↓	Anomaly-ShapeNet CD ↓	Anomaly-ShapeNet EMD ↓
w/o PE	0.0255	0.0133	0.0575	0.0276
with PE	0.0203	0.0110	0.0445	0.0228

Key Findings

1. Pose alignment is critical: Removing RANSAC causes O-AUROC to drop sharply from 0.900 to 0.711, demonstrating the necessity of global coarse registration
2. Dual role of positional encoding: Removing PE degrades P-AUROC by 11.4% (0.897→0.783) and also significantly reduces repair quality
3. Generalizability of PAM: PAM functions as a plug-and-play module that substantially improves the performance of various baseline methods
4. Advantage of continuous representation: PASDF outperforms all competing methods in 37/40 categories, whereas other methods exhibit inconsistent performance across categories

Highlights & Insights

1. Unified framework: This work is the first to unify 3D anomaly detection and repair within a continuous SDF representation, where detection and repair share a single learned shape representation
2. Pose-shape decoupling: PAM explicitly separates pose from shape, allowing the SDF network to focus on intrinsic geometric variations
3. Continuous vs. discrete: Replacing discrete voxel/point cloud/projection representations with continuous SDF avoids quantization artifacts and preserves fine geometric detail
4. Plug-and-play nature of PAM: PAM not only benefits PASDF but also markedly improves the performance of other methods, demonstrating strong generalizability

Limitations & Future Work

1. Computational cost: PAM registration is expensive when initial pose estimation is challenging; learning-based or hierarchical registration strategies could be explored
2. Single-class assumption: The current framework assumes a single category of normal objects; extending to multi-class detection would improve practical applicability
3. Sensitivity to input quality: Performance is affected by input point cloud quality, necessitating greater robustness to noise and outliers
4. Lack of contextual information: Integrating scene-level context could improve detection capability in complex real-world environments
5. Training efficiency: Training for 2000 epochs is time-consuming; more efficient training strategies are worth exploring

Related Work & Insights

• DeepSDF: The idea of parameterizing SDFs with neural networks originates from DeepSDF; PASDF creatively adapts this to anomaly detection
• ICP + RANSAC: The combination of these classical registration methods remains highly effective within the PAM pipeline
• Marching Cubes: This classical algorithm is employed in the repair stage to extract isosurfaces from the SDF
• Insight: Continuous representations (implicit functions) have a natural advantage over discrete representations in tasks requiring fine-grained geometric perception

Rating

• Novelty: ⭐⭐⭐⭐ — The unified SDF-based framework for 3D anomaly detection and repair is a novel and well-motivated contribution
• Technical Quality: ⭐⭐⭐⭐ — The method is thoroughly designed, ablations are comprehensive, and the generalizability of PAM is convincingly validated
• Experimental Thoroughness: ⭐⭐⭐⭐ — Two datasets, multiple baselines, detailed ablations, and qualitative results are provided
• Value: ⭐⭐⭐⭐ — The unified detection-and-repair framework has direct application value in manufacturing scenarios
• Overall Rating: ⭐⭐⭐⭐ (8/10)

Related Papers

• [ICCV 2025] PASDF: Bridging 3D Anomaly Localization and Repair via High-Quality Continuous Geometric Representation
• [ICCV 2025] SiM3D: Single-Instance Multiview Multimodal and Multisetup 3D Anomaly Detection Benchmark
• [ICCV 2025] RayletDF: Raylet Distance Fields for Generalizable 3D Surface Reconstruction from Point Clouds or Gaussians
• [ICCV 2025] Momentum-GS: Momentum Gaussian Self-Distillation for High-Quality Large Scene Reconstruction
• [ICCV 2025] DAP-MAE: Domain-Adaptive Point Cloud Masked Autoencoder for Effective Cross-Domain Learning