Zhu Yu

Shuhang Zheng, Yixuan Li, Zhu Yu et al.

Place recognition is an important technique for autonomous cars to achieve full autonomy since it can provide an initial guess to online localization algorithms. Although current methods based on images or point clouds have achieved satisfactory performance, localizing the images on a large-scale point cloud map remains a fairly unexplored problem. This cross-modal matching task is challenging due to the difficulty in extracting consistent descriptors from images and point clouds. In this paper, we propose the I2P-Rec method to solve the problem by transforming the cross-modal data into the same modality. Specifically, we leverage on the recent success of depth estimation networks to recover point clouds from images. We then project the point clouds into Bird's Eye View (BEV) images. Using the BEV image as an intermediate representation, we extract global features with a Convolutional Neural Network followed by a NetVLAD layer to perform matching. The experimental results evaluated on the KITTI dataset show that, with only a small set of training data, I2P-Rec achieves recall rates at Top-1\% over 80\% and 90\%, when localizing monocular and stereo images on point cloud maps, respectively. We further evaluate I2P-Rec on a 1 km trajectory dataset collected by an autonomous logistics car and show that I2P-Rec can generalize well to previously unseen environments.

Zhu Yu, Zhengyi Zhao, Runmin Zhang et al.

This work presents the Large Depth Completion Model (LDCM), a simple, effective, and robust framework for single-view metric depth estimation with sparse observations. Without relying on complex architectural designs, LDCM generates metric-accurate dense depth maps using a transformer. It outperforms existing approaches across diverse datasets and sparse observations. We achieve this from two key perspectives: (1) leveraging existing monocular foundation models to improve the quality of sparse depth inputs, and (2) reformulating training objectives to better capture geometric structure and metric consistency. Specifically, a Poisson-based depth initialization strategy is first introduced to generate a uniform coarse dense depth map from diverse sparse observations, providing a strong structural prior for the network. Regarding the training objective, we replace the conventional depth head with a point map head that regresses per-pixel 3D coordinates in camera space, enabling the model to directly learn the underlying 3D scene structure instead of performing pixel-wise depth map restoration. Moreover, this design eliminates the need for camera intrinsic parameters, allowing LDCM to naturally produce metric-scaled 3D point maps. Extensive experiments demonstrate that LDCM consistently outperforms state-of-the-art methods across multiple benchmarks and varying sparsity levels in both depth completion and point map estimation, showcasing its effectiveness and strong generalization to unseen data distributions.

Zhu Yu, Jingnan Gao, Runmin Zhang et al.

This work presents ViGeo, a feed-forward foundation model for recovering spatially dense and temporally consistent geometry from video sequences. Built upon a plain transformer architecture without task-specific architectural modifications, ViGeo supports streaming, full-sequence, and long-video inference within a unified model. The key design is dynamic chunking attention, which exposes the model to both bidirectional and causal temporal contexts during training and allows it to adapt its attention pattern at test time without retraining. To improve supervision quality, we further introduce a completion-based data refinement framework. This framework trains a video depth completion teacher that conditions on sparse and noisy annotations and exploits video/multi-view context to produce dense, temporally coherent, and geometrically reliable training targets. Beyond depth and point maps, ViGeo also predicts surface normals within the same framework. Trained solely on public datasets, ViGeo achieves state-of-the-art performance across online, offline, and long-video depth estimation, surface normal estimation, and video point map estimation.

Junchen Yu, Si-Yuan Cao, Runmin Zhang et al.

We propose a novel unsupervised cross-modal homography estimation learning framework, named Split Supervised Homography estimation Network (SSHNet). SSHNet reformulates the unsupervised cross-modal homography estimation into two supervised sub-problems, each addressed by its specialized network: a homography estimation network and a modality transfer network. To realize stable training, we introduce an effective split optimization strategy to train each network separately within its respective sub-problem. We also formulate an extra homography feature space supervision to enhance feature consistency, further boosting the estimation accuracy. Moreover, we employ a simple yet effective distillation training technique to reduce model parameters and improve cross-domain generalization ability while maintaining comparable performance. The training stability of SSHNet enables its cooperation with various homography estimation architectures. Experiments reveal that the SSHNet using IHN as homography estimation network, namely SSHNet-IHN, outperforms previous unsupervised approaches by a significant margin. Even compared to supervised approaches MHN and LocalTrans, SSHNet-IHN achieves 47.4% and 85.8% mean average corner errors (MACEs) reduction on the challenging OPT-SAR dataset.

Jiacheng Ying, Mushui Liu, Zhe Wu et al.

Blind face restoration has made great progress in producing high-quality and lifelike images. Yet it remains challenging to preserve the ID information especially when the degradation is heavy. Current reference-guided face restoration approaches either require face alignment or personalized test-tuning, which are unfaithful or time-consuming. In this paper, we propose a tuning-free method named RestorerID that incorporates ID preservation during face restoration. RestorerID is a diffusion model-based method that restores low-quality images with varying levels of degradation by using a single reference image. To achieve this, we propose a unified framework to combine the ID injection with the base blind face restoration model. In addition, we design a novel Face ID Rebalancing Adapter (FIR-Adapter) to tackle the problems of content unconsistency and contours misalignment that are caused by information conflicts between the low-quality input and reference image. Furthermore, by employing an Adaptive ID-Scale Adjusting strategy, RestorerID can produce superior restored images across various levels of degradation. Experimental results on the Celeb-Ref dataset and real-world scenarios demonstrate that RestorerID effectively delivers high-quality face restoration with ID preservation, achieving a superior performance compared to the test-tuning approaches and other reference-guided ones. The code of RestorerID is available at \url{https://github.com/YingJiacheng/RestorerID}.

Fuyi Zhang, Zhu Yu, Chunhao Li et al.

Radar has gained much attention in autonomous driving due to its accessibility and robustness. However, its standalone application for depth perception is constrained by issues of sparsity and noise. Radar-camera depth estimation offers a more promising complementary solution. Despite significant progress, current approaches fail to produce satisfactory dense depth maps, due to the unsatisfactory processing of the sparse and noisy radar data. They constrain the regions of interest for radar points in rigid rectangular regions, which may introduce unexpected errors and confusions. To address these issues, we develop a structure-aware strategy for radar depth enhancement, which provides more targeted regions of interest by leveraging the structural priors of RGB images. Furthermore, we design a Multi-Scale Structure Guided Network to enhance radar features and preserve detailed structures, achieving accurate and structure-detailed dense metric depth estimation. Building on these, we propose a structure-aware radar-camera depth estimation framework, named SA-RCD. Extensive experiments demonstrate that our SA-RCD achieves state-of-the-art performance on the nuScenes dataset. Our code will be available at https://github.com/FreyZhangYeh/SA-RCD.

Runmin Zhang, Zhu Yu, Si-Yuan Cao et al.

This work presents SGCDet, a novel multi-view indoor 3D object detection framework based on adaptive 3D volume construction. Unlike previous approaches that restrict the receptive field of voxels to fixed locations on images, we introduce a geometry and context aware aggregation module to integrate geometric and contextual information within adaptive regions in each image and dynamically adjust the contributions from different views, enhancing the representation capability of voxel features. Furthermore, we propose a sparse volume construction strategy that adaptively identifies and selects voxels with high occupancy probabilities for feature refinement, minimizing redundant computation in free space. Benefiting from the above designs, our framework achieves effective and efficient volume construction in an adaptive way. Better still, our network can be supervised using only 3D bounding boxes, eliminating the dependence on ground-truth scene geometry. Experimental results demonstrate that SGCDet achieves state-of-the-art performance on the ScanNet, ScanNet200 and ARKitScenes datasets. The source code is available at https://github.com/RM-Zhang/SGCDet.

Zhu Yu, Runmin Zhang, Jiacheng Ying et al.

Vision-based Semantic Scene Completion (SSC) has gained much attention due to its widespread applications in various 3D perception tasks. Existing sparse-to-dense approaches typically employ shared context-independent queries across various input images, which fails to capture distinctions among them as the focal regions of different inputs vary and may result in undirected feature aggregation of cross-attention. Additionally, the absence of depth information may lead to points projected onto the image plane sharing the same 2D position or similar sampling points in the feature map, resulting in depth ambiguity. In this paper, we present a novel context and geometry aware voxel transformer. It utilizes a context aware query generator to initialize context-dependent queries tailored to individual input images, effectively capturing their unique characteristics and aggregating information within the region of interest. Furthermore, it extend deformable cross-attention from 2D to 3D pixel space, enabling the differentiation of points with similar image coordinates based on their depth coordinates. Building upon this module, we introduce a neural network named CGFormer to achieve semantic scene completion. Simultaneously, CGFormer leverages multiple 3D representations (i.e., voxel and TPV) to boost the semantic and geometric representation abilities of the transformed 3D volume from both local and global perspectives. Experimental results demonstrate that CGFormer achieves state-of-the-art performance on the SemanticKITTI and SSCBench-KITTI-360 benchmarks, attaining a mIoU of 16.87 and 20.05, as well as an IoU of 45.99 and 48.07, respectively. Remarkably, CGFormer even outperforms approaches employing temporal images as inputs or much larger image backbone networks.

Runmin Zhang, Zhu Yu, Zehua Sheng et al.

Cross-spectral image guided denoising has shown its great potential in recovering clean images with rich details, such as using the near-infrared image to guide the denoising process of the visible one. To obtain such image pairs, a feasible and economical way is to employ a stereo system, which is widely used on mobile devices. Current works attempt to generate an aligned guidance image to handle the disparity between two images. However, due to occlusion, spectral differences and noise degradation, the aligned guidance image generally exists ghosting and artifacts, leading to an unsatisfactory denoised result. To address this issue, we propose a one-stage transformer-based architecture, named SGDFormer, for cross-spectral Stereo image Guided Denoising. The architecture integrates the correspondence modeling and feature fusion of stereo images into a unified network. Our transformer block contains a noise-robust cross-attention (NRCA) module and a spatially variant feature fusion (SVFF) module. The NRCA module captures the long-range correspondence of two images in a coarse-to-fine manner to alleviate the interference of noise. The SVFF module further enhances salient structures and suppresses harmful artifacts through dynamically selecting useful information. Thanks to the above design, our SGDFormer can restore artifact-free images with fine structures, and achieves state-of-the-art performance on various datasets. Additionally, our SGDFormer can be extended to handle other unaligned cross-model guided restoration tasks such as guided depth super-resolution.

Wuyang Li, Zhu Yu, Alexandre Alahi

3D semantic occupancy prediction aims to reconstruct the 3D geometry and semantics of the surrounding environment. With dense voxel labels, prior works typically formulate it as a dense segmentation task, independently classifying each voxel. However, this paradigm neglects critical instance-centric discriminability, leading to instance-level incompleteness and adjacent ambiguities. To address this, we highlight a free lunch of occupancy labels: the voxel-level class label implicitly provides insight at the instance level, which is overlooked by the community. Motivated by this observation, we first introduce a training-free Voxel-to-Instance (VoxNT) trick: a simple yet effective method that freely converts voxel-level class labels into instance-level offset labels. Building on this, we further propose VoxDet, an instance-centric framework that reformulates the voxel-level occupancy prediction as dense object detection by decoupling it into two sub-tasks: offset regression and semantic prediction. Specifically, based on the lifted 3D volume, VoxDet first uses (a) Spatially-decoupled Voxel Encoder to generate disentangled feature volumes for the two sub-tasks, which learn task-specific spatial deformation in the densely projected tri-perceptive space. Then, we deploy (b) Task-decoupled Dense Predictor to address this task via dense detection. Here, we first regress a 4D offset field to estimate distances (6 directions) between voxels and object borders in the voxel space. The regressed offsets are then used to guide the instance-level aggregation in the classification branch, achieving instance-aware prediction. Experiments show that VoxDet can be deployed on both camera and LiDAR input, jointly achieving state-of-the-art results on both benchmarks. VoxDet is not only highly efficient, but also achieves 63.0 IoU on the SemanticKITTI test set, ranking 1st on the online leaderboard.

Zhu Yu, Bowen Pang, Lizhe Liu et al.

We introduce LOcc, an effective and generalizable framework for open-vocabulary occupancy (OVO) prediction. Previous approaches typically supervise the networks through coarse voxel-to-text correspondences via image features as intermediates or noisy and sparse correspondences from voxel-based model-view projections. To alleviate the inaccurate supervision, we propose a semantic transitive labeling pipeline to generate dense and fine-grained 3D language occupancy ground truth. Our pipeline presents a feasible way to dig into the valuable semantic information of images, transferring text labels from images to LiDAR point clouds and ultimately to voxels, to establish precise voxel-to-text correspondences. By replacing the original prediction head of supervised occupancy models with a geometry head for binary occupancy states and a language head for language features, LOcc effectively uses the generated language ground truth to guide the learning of 3D language volume. Through extensive experiments, we demonstrate that our transitive semantic labeling pipeline can produce more accurate pseudo-labeled ground truth, diminishing labor-intensive human annotations. Additionally, we validate LOcc across various architectures, where all models consistently outperform state-of-the-art zero-shot occupancy prediction approaches on the Occ3D-nuScenes dataset.

Runmin Zhang, Jialiang Wang, Si-Yuan Cao et al.

This work presents DCFlow, a novel unsupervised cross-modal flow estimation framework that integrates a decoupled optimization strategy and a cross-modal consistency constraint. Unlike previous approaches that implicitly learn flow estimation solely from appearance similarity, we introduce a decoupled optimization strategy with task-specific supervision to address modality discrepancy and geometric misalignment distinctly. This is achieved by collaboratively training a modality transfer network and a flow estimation network. To enable reliable motion supervision without ground-truth flow, we propose a geometry-aware data synthesis pipeline combined with an outlier-robust loss. Additionally, we introduce a cross-modal consistency constraint to jointly optimize both networks, significantly improving flow prediction accuracy. For evaluation, we construct a comprehensive cross-modal flow benchmark by repurposing public datasets. Experimental results demonstrate that DCFlow can be integrated with various flow estimation networks and achieves state-of-the-art performance among unsupervised approaches.

Chenghao Zhang, Lun Luo, Si-Yuan Cao et al.

LiDAR-based global localization is an essential component of simultaneous localization and mapping (SLAM), which helps loop closure and re-localization. Current approaches rely on ground-truth poses obtained from GPS or SLAM odometry to supervise network training. Despite the great success of these supervised approaches, substantial cost and effort are required for high-precision ground-truth pose acquisition. In this work, we propose S-BEVLoc, a novel self-supervised framework based on bird's-eye view (BEV) for LiDAR global localization, which eliminates the need for ground-truth poses and is highly scalable. We construct training triplets from single BEV images by leveraging the known geographic distances between keypoint-centered BEV patches. Convolutional neural network (CNN) is used to extract local features, and NetVLAD is employed to aggregate global descriptors. Moreover, we introduce SoftCos loss to enhance learning from the generated triplets. Experimental results on the large-scale KITTI and NCLT datasets show that S-BEVLoc achieves state-of-the-art performance in place recognition, loop closure, and global localization tasks, while offering scalability that would require extra effort for supervised approaches.