CrossJEPA: Cross-Modal Joint-Embedding Predictive Architecture for Efficient 3D Representation Learning from 2D Images

• URL: http://arxiv.org/abs/2511.18424v1
• Date: Sun, 23 Nov 2025 12:40:04 GMT
• Title: CrossJEPA: Cross-Modal Joint-Embedding Predictive Architecture for Efficient 3D Representation Learning from 2D Images
• Authors: Avishka Perera, Kumal Hewagamage, Saeedha Nazar, Kavishka Abeywardana, Hasitha Gallella, Ranga Rodrigo, Mohamed Afham,
• Abstract summary: Cross-modal cross-modal learning has emerged to address the scarcity of large-scale 3D datasets in 3D representation learning.<n>We propose CrossJEPA, a simple Cross-modal Joint Embedding Predictive Architecture that harnesses the knowledge of an image foundation model and trains a predictor to infer embeddings of specific rendered 2D views from corresponding 3D point clouds.
• Score: 1.20952748584685
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Image-to-point cross-modal learning has emerged to address the scarcity of large-scale 3D datasets in 3D representation learning. However, current methods that leverage 2D data often result in large, slow-to-train models, making them computationally expensive and difficult to deploy in resource-constrained environments. The architecture design of such models is therefore critical, determining their performance, memory footprint, and compute efficiency. The Joint-embedding Predictive Architecture (JEPA) has gained wide popularity in self-supervised learning for its simplicity and efficiency, but has been under-explored in cross-modal settings, partly due to the misconception that masking is intrinsic to JEPA. In this light, we propose CrossJEPA, a simple Cross-modal Joint Embedding Predictive Architecture that harnesses the knowledge of an image foundation model and trains a predictor to infer embeddings of specific rendered 2D views from corresponding 3D point clouds, thereby introducing a JEPA-style pretraining strategy beyond masking. By conditioning the predictor on cross-domain projection information, CrossJEPA purifies the supervision signal from semantics exclusive to the target domain. We further exploit the frozen teacher design with a one-time target embedding caching mechanism, yielding amortized efficiency. CrossJEPA achieves a new state-of-the-art in linear probing on the synthetic ModelNet40 (94.2%) and the real-world ScanObjectNN (88.3%) benchmarks, using only 14.1M pretraining parameters (8.5M in the point encoder), and about 6 pretraining hours on a standard single GPU. These results position CrossJEPA as a performant, memory-efficient, and fast-to-train framework for 3D representation learning via knowledge distillation. We analyze CrossJEPA intuitively, theoretically, and empirically, and extensively ablate our design choices. Code will be made available.

Related papers

• BlendCLIP: Bridging Synthetic and Real Domains for Zero-Shot 3D Object Classification with Multimodal Pretraining [2.400704807305413]
  Zero-shot 3D object classification is crucial for real-world applications like autonomous driving.<n>It is often hindered by a significant domain gap between the synthetic data used for training and the sparse, noisy LiDAR scans encountered in the real-world.<n>We introduce BlendCLIP, a multimodal pretraining framework that bridges this synthetic-to-real gap by strategically combining the strengths of both domains.
  arXiv  Detail & Related papers  (2025-10-21T03:08:27Z)
• S3MOT: Monocular 3D Object Tracking with Selective State Space Model [3.5047603107971397]
  Multi-object tracking in 3D space is essential for advancing robotics and computer applications.<n>It remains a significant challenge in monocular setups due to the difficulty of mining 3D associations from 2D video streams.<n>We present three innovative techniques to enhance the fusion of heterogeneous cues for monocular 3D MOT.
  arXiv  Detail & Related papers  (2025-04-25T04:45:35Z)
• ALOcc: Adaptive Lifting-Based 3D Semantic Occupancy and Cost Volume-Based Flow Predictions [91.55655961014027]
  3D semantic occupancy and flow prediction are fundamental to understanding scene scene.<n>This paper proposes a vision-based framework with three targeted improvements.<n>Our purely convolutional architecture establishes new SOTA performance on multiple benchmarks for both semantic occupancy and joint semantic-flow prediction.
  arXiv  Detail & Related papers  (2024-11-12T11:32:56Z)
• PillarNeXt: Rethinking Network Designs for 3D Object Detection in LiDAR Point Clouds [29.15589024703907]
  In this paper, we revisit the local point aggregators from the perspective of allocating computational resources. We find that the simplest pillar based models perform surprisingly well considering both accuracy and latency. Our results challenge the common intuition that the detailed geometry modeling is essential to achieve high performance for 3D object detection.
  arXiv  Detail & Related papers  (2023-05-08T17:59:14Z)
• Exploring Object-Centric Temporal Modeling for Efficient Multi-View 3D Object Detection [20.161887223481994]
  We propose a long-sequence modeling framework, named StreamPETR, for multi-view 3D object detection. StreamPETR achieves significant performance improvements only with negligible cost, compared to the single-frame baseline. The lightweight version realizes 45.0% mAP and 31.7 FPS, outperforming the state-of-the-art method (SOLOFusion) by 2.3% mAP and 1.8x faster FPS.
  arXiv  Detail & Related papers  (2023-03-21T15:19:20Z)
• Efficient Few-Shot Object Detection via Knowledge Inheritance [62.36414544915032]
  Few-shot object detection (FSOD) aims at learning a generic detector that can adapt to unseen tasks with scarce training samples. We present an efficient pretrain-transfer framework (PTF) baseline with no computational increment. We also propose an adaptive length re-scaling (ALR) strategy to alleviate the vector length inconsistency between the predicted novel weights and the pretrained base weights.
  arXiv  Detail & Related papers  (2022-03-23T06:24:31Z)
• GDRNPP: A Geometry-guided and Fully Learning-based Object Pose Estimator [51.89441403642665]
  6D pose estimation of rigid objects is a long-standing and challenging task in computer vision.<n>Recently, the emergence of deep learning reveals the potential of Convolutional Neural Networks (CNNs) to predict reliable 6D poses.<n>This paper introduces a fully learning-based object pose estimator.
  arXiv  Detail & Related papers  (2021-02-24T09:11:31Z)
• Self-Supervised Pretraining of 3D Features on any Point-Cloud [40.26575888582241]
  We present a simple self-supervised pertaining method that can work with any 3D data without 3D registration. We evaluate our models on 9 benchmarks for object detection, semantic segmentation, and object classification, where they achieve state-of-the-art results and can outperform supervised pretraining.
  arXiv  Detail & Related papers  (2021-01-07T18:55:21Z)
• Reinforced Axial Refinement Network for Monocular 3D Object Detection [160.34246529816085]
  Monocular 3D object detection aims to extract the 3D position and properties of objects from a 2D input image. Conventional approaches sample 3D bounding boxes from the space and infer the relationship between the target object and each of them, however, the probability of effective samples is relatively small in the 3D space. We propose to start with an initial prediction and refine it gradually towards the ground truth, with only one 3d parameter changed in each step. This requires designing a policy which gets a reward after several steps, and thus we adopt reinforcement learning to optimize it.
  arXiv  Detail & Related papers  (2020-08-31T17:10:48Z)
• 2nd Place Scheme on Action Recognition Track of ECCV 2020 VIPriors Challenges: An Efficient Optical Flow Stream Guided Framework [57.847010327319964]
  We propose a data-efficient framework that can train the model from scratch on small datasets. Specifically, by introducing a 3D central difference convolution operation, we proposed a novel C3D neural network-based two-stream framework. It is proved that our method can achieve a promising result even without a pre-trained model on large scale datasets.
  arXiv  Detail & Related papers  (2020-08-10T09:50:28Z)
• Inference Stage Optimization for Cross-scenario 3D Human Pose Estimation [97.93687743378106]
  Existing 3D pose estimation models suffer performance drop when applying to new scenarios with unseen poses. We propose a novel framework, Inference Stage Optimization (ISO), for improving the generalizability of 3D pose models. Remarkably, it yields new state-of-the-art of 83.6% 3D PCK on MPI-INF-3DHP, improving upon the previous best result by 9.7%.
  arXiv  Detail & Related papers  (2020-07-04T09:45:18Z)
• 3DSSD: Point-based 3D Single Stage Object Detector [61.67928229961813]
  We present a point-based 3D single stage object detector, named 3DSSD, achieving a good balance between accuracy and efficiency. Our method outperforms all state-of-the-art voxel-based single stage methods by a large margin, and has comparable performance to two stage point-based methods as well.
  arXiv  Detail & Related papers  (2020-02-24T12:01:58Z)

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.