Foundation4D is a unified feed‑forward model that performs geometry estimation, 3D detection, semantic occupancy, and Gaussian reconstruction from multi‑camera driving video. It uses a shared backbone with factorized spatiotemporal attention to merge observations across cameras and time, and a video inference pipeline that refines camera trajectories via pose graph optimization and assembles object tracks. The authors also provide a data annotation pipeline and release the PAI‑Foundation4D benchmark, achieving state‑of‑the‑art results on depth, detection, and occupancy metrics and demonstrating that co‑training detection with geometry and occupancy improves performance.
arXiv:2608.13147v2 Announce Type: replace Abstract: Camera-based autonomous driving perception requires a shared representation that preserves metric 3D structure across synchronized multi-camera str...
TAPVid-MV is a new benchmark for tracking any point in 3D across multiple synchronized camera views. It comprises 284 sequences, 1,142 calibrated camera streams, and 109,769 point tracks, covering indoor and outdoor domains and derived from various modalities such as depth, LiDAR, SLAM, and simulation. The dataset is visually verified, and evaluation shows that current multi‑view trackers do not consistently outperform monocular trackers, highlighting geometry recovery as a key bottleneck.
The paper introduces a biologically inspired framework that learns object‑centric visual representations from raw videos without human annotations or camera calibration. By using motion boundaries detected via optical flow and clustering to create pseudo‑instance masks, the method supervises a single‑image encoder with pixel‑level pairwise metric learning. Training on 195 million pseudo‑labeled frames and expanding to 421 million frames through Motion‑Verified Self‑Training, the approach yields Swin‑based encoders that outperform or match supervised and self‑supervised baselines on tasks such as monocular depth estimation, 3D object detection, 3D occupancy prediction, and end‑to‑end planning.
PURE is an unlabeled post‑training refinement framework that enhances feed‑forward 3D reconstruction models by improving depth and camera estimation without ground‑truth supervision. It distills structural knowledge from state‑of‑the‑art 2D priors into 4D architectures, using relative depth constraints, affine‑invariant geometry distillation, and pose refinement via feature correspondences. Experiments on synthetic and real‑world dynamic benchmarks show consistent improvements across multiple state‑of‑the‑art models, effectively mitigating domain shifts and unlocking their architectural limits in challenging in‑the‑wild scenes.
arXiv:2609.22896v1 Announce Type: new Abstract: Autonomous vehicles operating in open-world scenarios are inevitably confronted with previously unknown objects, such as exotic animals or loose cargo....
arXiv:2512. 16919v2 Announce Type: replace-cross Abstract: Perceiving and reconstructing 3D scene geometry from visual inputs is crucial for autonomous driving.
The paper introduces LiDAR‑SAM2, a framework that converts the 2D video foundation model SAM2 into a scalable source of supervision for 4D LiDAR data. By projecting SAM2 video masks into multi‑view LiDAR space and aggregating them temporally, the method automatically generates temporally coherent LiDAR labels without human annotation. Experiments on SemanticKITTI show that these automatically produced semantic and panoptic labels achieve quality close to full human annotation, enabling models trained on them to approach the performance of fully supervised systems.
PanoBench is a new camera‑centric benchmark for panoptic occupancy prediction that unifies 3D geometry, semantics, and instance identities across synthetic and real‑world driving scenes. It combines three datasets—PanoBench‑CarlaOcc, PanoBench‑nuScenes, and PanoBench‑KITTI360—through canonical interfaces and unified evaluation protocols, providing multi‑resolution annotations with persistent instance identities. The benchmark introduces a novel reconstruction pipeline for real‑world data, a unified valid‑mask protocol, and analyses of observation fidelity, geometric completeness, spatial continuity, and temporal consistency, while also benchmarking methods on semantic, panoptic, and 4D occupancy forecasting. whyItMatters":"PanoBench addresses the scarcity of instance‑aware annotations in panoptic occupancy research by offering comprehensive, high‑quality annotations and evaluation protocols that enable more robust development and comparison of autonomous driving perception methods."
arXiv:2609.01172v1 Announce Type: new Abstract: Monocular depth estimation has long stood as a fundamental challenge in computer vision, enabling a wide range of applications including 3D reconstruct...
arXiv:2603.12064v3 Announce Type: replace Abstract: We address the challenging problem of dense dynamic scene reconstruction and camera pose estimation from multiple freely moving cameras -- a settin...
Dyna3 is a training‑free framework that extends the depth foundation model DA3 to perform 4D dynamic scene reconstruction without fine‑tuning. By leveraging DA3’s cross‑view features and a best‑match search, it distinguishes static surfaces from moving objects, and uses vision‑language models to generate semantic prompts for SAM 3 to achieve precise instance‑level segmentation. Experiments on four datasets show Dyna3 outperforms correspondence‑trained methods, improving dynamic object segmentation by +5.5 pp, speeding pose estimation 13×, and reducing memory usage 4–8×.
arXiv:2606. 24353v1 Announce Type: cross Abstract: Bird's-eye view (BEV) perception fuses multi-camera images into a unified top-down representation for autonomous driving.