Mitigating Performance Discrepancy in Cross-Domain 3D Class-Incremental Learning

• URL: http://arxiv.org/abs/2609.04860v1
• Date: Fri, 04 Sep 2026 08:20:00 GMT
• Title: Mitigating Performance Discrepancy in Cross-Domain 3D Class-Incremental Learning
• Abstract summary: 3D perception plays a crucial role in real-world applications such as autonomous driving, robotics, and AR/VR.<n>Unlike 2D images, 3D point clouds are inherently heterogeneous.<n>The degree of performance degradation can vary substantially across domains, a phenomenon we term performance discrepancy.
• Score: 9.38434232095379
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 3D perception plays a crucial role in real-world applications such as autonomous driving, robotics, and AR/VR. In practical scenarios, 3D perception models need to continually adapt to newly emerging 3D object categories, making class-incremental learning (CIL) particularly important. However, unlike 2D images, 3D point clouds are inherently heterogeneous: objects from the same class may not only come from the clean CAD domain, but also from RGB-D camera scans of varying quality, video reconstructions, or even corrupted observations. We discover that such heterogeneity introduces a new challenge beyond catastrophic forgetting: the degree of performance degradation can vary substantially across domains, a phenomenon we term performance discrepancy. To investigate this problem, we establish the Domain3D-CIL training and evaluation protocol, which contains point cloud categories from heterogeneous domains. We further adapt a wide range of mainstream CIL methods to the 3D modality. The results demonstrate that this performance discrepancy consistently appears across these baselines. To mitigate this issue, we introduce PolyMem, an exemplar-free approach that implicitly models rich high-order statistics of the feature distribution to enhance cross-domain robustness. Experiments demonstrate that our method effectively alleviates the performance discrepancy while improving the model's performance across domains. Code will be made publicly available upon acceptance.

Related papers

• UniDA3D: A Unified Domain-Adaptive Framework for Multi-View 3D Object Detection [59.35818507180513]
  Camera-only 3D object detection is critical for autonomous driving, offering a cost-effective alternative to LiDAR based methods.<n>Existing methods often suffer significant performance degradation under complex environmental conditions such as nighttime, fog, and rain.<n>We propose UniDA3D, a unified domain-adaptive multi-view 3D object detector designed for robust perception under diverse adverse conditions.
  arXiv  Detail & Related papers  (2026-03-30T03:35:32Z)
• Topology-Aware Modeling for Unsupervised Simulation-to-Reality Point Cloud Recognition [63.55828203989405]
  We introduce a novel Topology-Aware Modeling (TAM) framework for Sim2Real UDA on object point clouds.<n>Our approach mitigates the domain gap by leveraging global spatial topology, characterized by low-level, high-frequency 3D structures.<n>We propose an advanced self-training strategy that combines cross-domain contrastive learning with self-training.
  arXiv  Detail & Related papers  (2025-06-26T11:53:59Z)
• Revisiting Cross-Domain Problem for LiDAR-based 3D Object Detection [5.149095033945412]
  We deeply analyze the cross-domain performance of the state-of-the-art models. We observe that most models will overfit the training domains and it is challenging to adapt them to other domains directly. We propose additional evaluation metrics -- the side-view and front-view AP -- to better analyze the core issues of the methods' heavy drops in accuracy levels.
  arXiv  Detail & Related papers  (2024-08-22T19:52:44Z)
• Source-Free and Image-Only Unsupervised Domain Adaptation for Category Level Object Pose Estimation [18.011044932979143]
  3DUDA is a method capable of adapting to a nuisance-ridden target domain without 3D or depth data. We represent object categories as simple cuboid meshes, and harness a generative model of neural feature activations. We show that our method simulates fine-tuning on a global pseudo-labeled dataset under mild assumptions.
  arXiv  Detail & Related papers  (2024-01-19T17:48:05Z)
• FILP-3D: Enhancing 3D Few-shot Class-incremental Learning with Pre-trained Vision-Language Models [59.13757801286343]
  Few-shot class-incremental learning aims to mitigate the catastrophic forgetting issue when a model is incrementally trained on limited data.<n>We introduce the FILP-3D framework with two novel components: the Redundant Feature Eliminator (RFE) for feature space misalignment and the Spatial Noise Compensator (SNC) for significant noise.
  arXiv  Detail & Related papers  (2023-12-28T14:52:07Z)
• 3D Adversarial Augmentations for Robust Out-of-Domain Predictions [115.74319739738571]
  We focus on improving the generalization to out-of-domain data. We learn a set of vectors that deform the objects in an adversarial fashion. We perform adversarial augmentation by applying the learned sample-independent vectors to the available objects when training a model.
  arXiv  Detail & Related papers  (2023-08-29T17:58:55Z)
• Revisiting Domain-Adaptive 3D Object Detection by Reliable, Diverse and Class-balanced Pseudo-Labeling [38.07637524378327]
  Unsupervised domain adaptation (DA) with the aid of pseudo labeling techniques has emerged as a crucial approach for domain-adaptive 3D object detection. Existing DA methods suffer from a substantial drop in performance when applied to a multi-class training setting. We propose a novel ReDB framework tailored for learning to detect all classes at once.
  arXiv  Detail & Related papers  (2023-07-16T04:34:11Z)
• DA-CIL: Towards Domain Adaptive Class-Incremental 3D Object Detection [2.207918236777924]
  We propose a novel 3D domain adaptive class-incremental object detection framework, DA-CIL. We design a novel dual-domain copy-paste augmentation method to construct multiple augmented domains for diversifying training distributions. Experiments on various datasets demonstrate the effectiveness of the proposed method over baselines.
  arXiv  Detail & Related papers  (2022-12-05T06:45:27Z)
• Towards Model Generalization for Monocular 3D Object Detection [57.25828870799331]
  We present an effective unified camera-generalized paradigm (CGP) for Mono3D object detection. We also propose the 2D-3D geometry-consistent object scaling strategy (GCOS) to bridge the gap via an instance-level augment. Our method called DGMono3D achieves remarkable performance on all evaluated datasets and surpasses the SoTA unsupervised domain adaptation scheme.
  arXiv  Detail & Related papers  (2022-05-23T23:05:07Z)
• Unsupervised Domain Adaptation for Monocular 3D Object Detection via Self-Training [57.25828870799331]
  We propose STMono3D, a new self-teaching framework for unsupervised domain adaptation on Mono3D. We develop a teacher-student paradigm to generate adaptive pseudo labels on the target domain. STMono3D achieves remarkable performance on all evaluated datasets and even surpasses fully supervised results on the KITTI 3D object detection dataset.
  arXiv  Detail & Related papers  (2022-04-25T12:23:07Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.