Lijun Zhao

Kun Dai, Tao Xie, Ke Wang et al.

Local feature matching is an essential component in many visual applications. In this work, we propose OAMatcher, a Tranformer-based detector-free method that imitates humans behavior to generate dense and accurate matches. Firstly, OAMatcher predicts overlapping areas to promote effective and clean global context aggregation, with the key insight that humans focus on the overlapping areas instead of the entire images after multiple observations when matching keypoints in image pairs. Technically, we first perform global information integration across all keypoints to imitate the humans behavior of observing the entire images at the beginning of feature matching. Then, we propose Overlapping Areas Prediction Module (OAPM) to capture the keypoints in co-visible regions and conduct feature enhancement among them to simulate that humans transit the focus regions from the entire images to overlapping regions, hence realizeing effective information exchange without the interference coming from the keypoints in non overlapping areas. Besides, since humans tend to leverage probability to determine whether the match labels are correct or not, we propose a Match Labels Weight Strategy (MLWS) to generate the coefficients used to appraise the reliability of the ground-truth match labels, while alleviating the influence of measurement noise coming from the data. Moreover, we integrate depth-wise convolution into Tranformer encoder layers to ensure OAMatcher extracts local and global feature representation concurrently. Comprehensive experiments demonstrate that OAMatcher outperforms the state-of-the-art methods on several benchmarks, while exhibiting excellent robustness to extreme appearance variants. The source code is available at https://github.com/DK-HU/OAMatcher.

Tao Xie, Kun Dai, Siyi Lu et al.

In this work, we seek to predict camera poses across scenes with a multi-task learning manner, where we view the localization of each scene as a new task. We propose OFVL-MS, a unified framework that dispenses with the traditional practice of training a model for each individual scene and relieves gradient conflict induced by optimizing multiple scenes collectively, enabling efficient storage yet precise visual localization for all scenes. Technically, in the forward pass of OFVL-MS, we design a layer-adaptive sharing policy with a learnable score for each layer to automatically determine whether the layer is shared or not. Such sharing policy empowers us to acquire task-shared parameters for a reduction of storage cost and task-specific parameters for learning scene-related features to alleviate gradient conflict. In the backward pass of OFVL-MS, we introduce a gradient normalization algorithm that homogenizes the gradient magnitude of the task-shared parameters so that all tasks converge at the same pace. Furthermore, a sparse penalty loss is applied on the learnable scores to facilitate parameter sharing for all tasks without performance degradation. We conduct comprehensive experiments on multiple benchmarks and our new released indoor dataset LIVL, showing that OFVL-MS families significantly outperform the state-of-the-arts with fewer parameters. We also verify that OFVL-MS can generalize to a new scene with much few parameters while gaining superior localization performance.

Tao Xie, Kun Dai, Ke Wang et al.

Local feature matching between images remains a challenging task, especially in the presence of significant appearance variations, e.g., extreme viewpoint changes. In this work, we propose DeepMatcher, a deep Transformer-based network built upon our investigation of local feature matching in detector-free methods. The key insight is that local feature matcher with deep layers can capture more human-intuitive and simpler-to-match features. Based on this, we propose a Slimming Transformer (SlimFormer) dedicated for DeepMatcher, which leverages vector-based attention to model relevance among all keypoints and achieves long-range context aggregation in an efficient and effective manner. A relative position encoding is applied to each SlimFormer so as to explicitly disclose relative distance information, further improving the representation of keypoints. A layer-scale strategy is also employed in each SlimFormer to enable the network to assimilate message exchange from the residual block adaptively, thus allowing it to simulate the human behaviour that humans can acquire different matching cues each time they scan an image pair. To facilitate a better adaption of the SlimFormer, we introduce a Feature Transition Module (FTM) to ensure a smooth transition in feature scopes with different receptive fields. By interleaving the self- and cross-SlimFormer multiple times, DeepMatcher can easily establish pixel-wise dense matches at coarse level. Finally, we perceive the match refinement as a combination of classification and regression problems and design Fine Matches Module to predict confidence and offset concurrently, thereby generating robust and accurate matches. Experimentally, we show that DeepMatcher significantly outperforms the state-of-the-art methods on several benchmarks, demonstrating the superior matching capability of DeepMatcher.

Hao Jing, Anhong Wang, Lijun Zhao et al.

In autonomous driving, LiDAR sensors are vital for acquiring 3D point clouds, providing reliable geometric information. However, traditional sampling methods of preprocessing often ignore semantic features, leading to detail loss and ground point interference in 3D object detection. To address this, we propose a multi-branch two-stage 3D object detection framework using a Semantic-aware Multi-branch Sampling (SMS) module and multi-view consistency constraints. The SMS module includes random sampling, Density Equalization Sampling (DES) for enhancing distant objects, and Ground Abandonment Sampling (GAS) to focus on non-ground points. The sampled multi-view points are processed through a Consistent KeyPoint Selection (CKPS) module to generate consistent keypoint masks for efficient proposal sampling. The first-stage detector uses multi-branch parallel learning with multi-view consistency loss for feature aggregation, while the second-stage detector fuses multi-view data through a Multi-View Fusion Pooling (MVFP) module to precisely predict 3D objects. The experimental results on the KITTI dataset and Waymo Open Dataset show that our method achieves excellent detection performance improvement for a variety of backbones, especially for low-performance backbones with the simple network structures.

Qihao Sun, Jiarun Liu, Ziqian Ni et al.

Accurate metric depth is critical for autonomous driving perception and simulation, yet current approaches struggle to achieve high metric accuracy, multi-view and temporal consistency, and cross-domain generalization. To address these challenges, we present DriveMVS, a novel multi-view stereo framework that reconciles these competing objectives through two key insights: (1) Sparse but metrically accurate LiDAR observations can serve as geometric prompts to anchor depth estimation in absolute scale, and (2) deep fusion of diverse cues is essential for resolving ambiguities and enhancing robustness, while a spatio-temporal decoder ensures consistency across frames. Built upon these principles, DriveMVS embeds the LiDAR prompt in two ways: as a hard geometric prior that anchors the cost volume, and as soft feature-wise guidance fused by a triple-cue combiner. Regarding temporal consistency, DriveMVS employs a spatio-temporal decoder that jointly leverages geometric cues from the MVS cost volume and temporal context from neighboring frames. Experiments show that DriveMVS achieves state-of-the-art performance on multiple benchmarks, excelling in metric accuracy, temporal stability, and zero-shot cross-domain transfer, demonstrating its practical value for scalable, reliable autonomous driving systems.

Xiaoyu Li, Dedong Liu, Yitao Wu et al.

3D Multi-Object Tracking (MOT) captures stable and comprehensive motion states of surrounding obstacles, essential for robotic perception. However, current 3D trackers face issues with accuracy and latency consistency. In this paper, we propose Fast-Poly, a fast and effective filter-based method for 3D MOT. Building upon our previous work Poly-MOT, Fast-Poly addresses object rotational anisotropy in 3D space, enhances local computation densification, and leverages parallelization technique, improving inference speed and precision. Fast-Poly is extensively tested on two large-scale tracking benchmarks with Python implementation. On the nuScenes dataset, Fast-Poly achieves new state-of-the-art performance with 75.8% AMOTA among all methods and can run at 34.2 FPS on a personal CPU. On the Waymo dataset, Fast-Poly exhibits competitive accuracy with 63.6% MOTA and impressive inference speed (35.5 FPS). The source code is publicly available at https://github.com/lixiaoyu2000/FastPoly.

Xiaoyu Li, Peidong Li, Lijun Zhao et al.

3D Multi-Object Tracking (MOT) obtains significant performance improvements with the rapid advancements in 3D object detection, particularly in cost-effective multi-camera setups. However, the prevalent end-to-end training approach for multi-camera trackers results in detector-specific models, limiting their versatility. Moreover, current generic trackers overlook the unique features of multi-camera detectors, i.e., the unreliability of motion observations and the feasibility of visual information. To address these challenges, we propose RockTrack, a 3D MOT method for multi-camera detectors. Following the Tracking-By-Detection framework, RockTrack is compatible with various off-the-shelf detectors. RockTrack incorporates a confidence-guided preprocessing module to extract reliable motion and image observations from distinct representation spaces from a single detector. These observations are then fused in an association module that leverages geometric and appearance cues to minimize mismatches. The resulting matches are propagated through a staged estimation process, forming the basis for heuristic noise modeling. Additionally, we introduce a novel appearance similarity metric for explicitly characterizing object affinities in multi-camera settings. RockTrack achieves state-of-the-art performance on the nuScenes vision-only tracking leaderboard with 59.1% AMOTA while demonstrating impressive computational efficiency.

Yingzhao Li, Yan Li, Shixiong Tian et al.

This paper introduces MipSLAM, a frequency-aware 3D Gaussian Splatting (3DGS) SLAM framework capable of high-fidelity anti-aliased novel view synthesis and robust pose estimation under varying camera configurations. Existing 3DGS-based SLAM systems often suffer from aliasing artifacts and trajectory drift due to inadequate filtering and purely spatial optimization. To overcome these limitations, we propose an Elliptical Adaptive Anti-aliasing (EAA) algorithm that approximates Gaussian contributions via geometry-aware numerical integration, avoiding costly analytic computation. Furthermore, we present a Spectral-Aware Pose Graph Optimization (SA-PGO) module that reformulates trajectory estimation in the frequency domain, effectively suppressing high-frequency noise and drift through graph Laplacian analysis. A novel local frequency-domain perceptual loss is also introduced to enhance fine-grained geometric detail recovery. Extensive evaluations on Replica and TUM datasets demonstrate that MipSLAM achieves state-of-the-art rendering quality and localization accuracy across multiple resolutions while maintaining real-time capability. Code is available at https://github.com/yzli1998/MipSLAM.

Xiaoyu Li, Peidong Li, Xian Wu et al.

Spatio-temporal alignment is crucial for temporal modeling of end-to-end (E2E) perception in autonomous driving (AD), providing valuable structural and textural prior information. Existing methods typically rely on the attention mechanism to align objects across frames, simplifying the motion model with a unified explicit physical model (constant velocity, etc.). These approaches prefer semantic features for implicit alignment, challenging the importance of explicit motion modeling in the traditional perception paradigm. However, variations in motion states and object features across categories and frames render this alignment suboptimal. To address this, we propose HAT, a spatio-temporal alignment module that allows each object to adaptively decode the optimal alignment proposal from multiple hypotheses without direct supervision. Specifically, HAT first utilizes multiple explicit motion models to generate spatial anchors and motion-aware feature proposals for historical instances. It then performs multi-hypothesis decoding by incorporating semantic and motion cues embedded in cached object queries, ultimately providing the optimal alignment proposal for the target frame. On nuScenes, HAT consistently improves 3D temporal detectors and trackers across diverse baselines. It achieves state-of-the-art tracking results with 46.0% AMOTA on the test set when paired with the DETR3D detector. In an object-centric E2E AD method, HAT enhances perception accuracy (+1.3% mAP, +3.1% AMOTA) and reduces the collision rate by 32%. When semantics are corrupted (nuScenes-C), the enhancement of motion modeling by HAT enables more robust perception and planning in the E2E AD.

Xian Wu, Yitao Wu, Xiaoyu Li et al.

LiDAR-camera 3D multi-object tracking (MOT) combines rich visual semantics with accurate depth cues to improve trajectory consistency and tracking reliability. In practice, however, LiDAR and cameras operate at different sampling rates. To maintain temporal alignment, existing data pipelines usually synchronize heterogeneous sensor streams and annotate them at a reduced shared frequency, forcing most prior methods to perform spatial fusion only at synchronized timestamps through projection-based or learnable cross-sensor association. As a result, abundant asynchronous observations remain underexploited, despite their potential to support more frequent association and more robust trajectory estimation over short temporal intervals. To address this limitation, we propose Fusion-Poly, a spatial-temporal fusion framework for 3D MOT that integrates asynchronous LiDAR and camera data. Fusion-Poly associates trajectories with multi-modal observations at synchronized timestamps and with single-modal observations at asynchronous timestamps, enabling higher-frequency updates of motion and existence states. The framework contains three key components: a frequency-aware cascade matching module that adapts to synchronized and asynchronous frames according to available detection modalities; a frequency-aware trajectory estimation module that maintains trajectories through high-frequency motion prediction, differential updates, and confidence-calibrated lifecycle management; and a full-state observation alignment module that improves cross-modal consistency at synchronized timestamps by optimizing image-projection errors. On the nuScenes test set, Fusion-Poly achieves 76.5% AMOTA, establishing a new state of the art among tracking-by-detection 3D MOT methods. Extensive ablation studies further validate the effectiveness of each component. Code will be released.

Lijun Zhao, Huihui Bai, Anhong Wang et al.

In this paper, we introduce a deep multiple description coding (MDC) framework optimized by minimizing multiple description (MD) compressive loss. First, MD multi-scale-dilated encoder network generates multiple description tensors, which are discretized by scalar quantizers, while these quantized tensors are decompressed by MD cascaded-ResBlock decoder networks. To greatly reduce the total amount of artificial neural network parameters, an auto-encoder network composed of these two types of network is designed as a symmetrical parameter sharing structure. Second, this autoencoder network and a pair of scalar quantizers are simultaneously learned in an end-to-end self-supervised way. Third, considering the variation in the image spatial distribution, each scalar quantizer is accompanied by an importance-indicator map to generate MD tensors, rather than using direct quantization. Fourth, we introduce the multiple description structural similarity distance loss, which implicitly regularizes the diversified multiple description generations, to explicitly supervise multiple description diversified decoding in addition to MD reconstruction loss. Finally, we demonstrate that our MDC framework performs better than several state-of-the-art MDC approaches regarding image coding efficiency when tested on several commonly available datasets.

Lijun Zhao, Jinjing Zhang, Fan Zhang et al.

Most deep image smoothing operators are always trained repetitively when different explicit structure-texture pairs are employed as label images for each algorithm configured with different parameters. This kind of training strategy often takes a long time and spends equipment resources in a costly manner. To address this challenging issue, we generalize continuous network interpolation as a more powerful model generation tool, and then propose a simple yet effective model generation strategy to form a sequence of models that only requires a set of specific-effect label images. To precisely learn image smoothing operators, we present a double-state aggregation (DSA) module, which can be easily inserted into most of current network architecture. Based on this module, we design a double-state aggregation neural network structure with a local feature aggregation block and a nonlocal feature aggregation block to obtain operators with large expression capacity. Through the evaluation of many objective and visual experimental results, we show that the proposed method is capable of producing a series of continuous models and achieves better performance than that of several state-of-the-art methods for image smoothing.

Xian Wei, Hao Shen, Yuanxiang Li et al.

There are some inadequacies in the language description of this paper that require further improvement. This paper is based on a revision of a conference paper. It is now necessary to further explain the difference between the contributions of the two papers.

Lijun Zhao, Huihui Bai, Anhong Wang et al.

In this paper, we propose a deep multiple description coding framework, whose quantizers are adaptively learned via the minimization of multiple description compressive loss. Firstly, our framework is built upon auto-encoder networks, which have multiple description multi-scale dilated encoder network and multiple description decoder networks. Secondly, two entropy estimation networks are learned to estimate the informative amounts of the quantized tensors, which can further supervise the learning of multiple description encoder network to represent the input image delicately. Thirdly, a pair of scalar quantizers accompanied by two importance-indicator maps is automatically learned in an end-to-end self-supervised way. Finally, multiple description structural dissimilarity distance loss is imposed on multiple description decoded images in pixel domain for diversified multiple description generations rather than on feature tensors in feature domain, in addition to multiple description reconstruction loss. Through testing on two commonly used datasets, it is verified that our method is beyond several state-of-the-art multiple description coding approaches in terms of coding efficiency.

Dongao Ma, Ping Tang, Lijun Zhao

Lack of annotated samples greatly restrains the direct application of deep learning in remote sensing image scene classification. Although researches have been done to tackle this issue by data augmentation with various image transformation operations, they are still limited in quantity and diversity. Recently, the advent of the unsupervised learning based generative adversarial networks (GANs) bring us a new way to generate augmented samples. However, such GAN-generated samples are currently only served for training GANs model itself and for improving the performance of the discriminator in GANs internally (in vivo). It becomes a question of serious doubt whether the GAN-generated samples can help better improve the scene classification performance of other deep learning networks (in vitro), compared with the widely used transformed samples. To answer this question, this paper proposes a SiftingGAN approach to generate more numerous, more diverse and more authentic labeled samples for data augmentation. SiftingGAN extends traditional GAN framework with an Online-Output method for sample generation, a Generative-Model-Sifting method for model sifting, and a Labeled-Sample-Discriminating method for sample sifting. Experiments on the well-known AID dataset demonstrate that the proposed SiftingGAN method can not only effectively improve the performance of the scene classification baseline that is achieved without data augmentation, but also significantly excels the comparison methods based on traditional geometric/radiometric transformation operations.

Lijun Zhao, Huihui Bai, Anhong Wang et al.

In this paper, we propose an image re-sampling compression method by learning virtual codec network (VCN) to resolve the non-differentiable problem of quantization function for image compression. Here, the image re-sampling not only refers to image full-resolution re-sampling but also low-resolution re-sampling. We generalize this method for standard-compliant image compression (SCIC) framework and deep neural networks based compression (DNNC) framework. Specifically, an input image is measured by re-sampling network (RSN) network to get re-sampled vectors. Then, these vectors are directly quantized in the feature space in SCIC, or discrete cosine transform coefficients of these vectors are quantized to further improve coding efficiency in DNNC. At the encoder, the quantized vectors or coefficients are losslessly compressed by arithmetic coding. At the receiver, the decoded vectors are utilized to restore input image by image decoder network (IDN). In order to train RSN network and IDN network together in an end-to-end fashion, our VCN network intimates projection from the re-sampled vectors to the IDN-decoded image. As a result, gradients from IDN network to RSN network can be approximated by VCN network's gradient. Because dimension reduction can be further achieved by quantization in some dimensional space after image re-sampling within auto-encoder architecture, we can well initialize our networks from pre-trained auto-encoder networks. Through extensive experiments and analysis, it is verified that the proposed method has more effectiveness and versatility than many state-of-the-art approaches.

Lijun Zhao, Huihui Bai, Feng Li et al.

In this paper, we propose an end-to-end mixed-resolution image compression framework with convolutional neural networks. Firstly, given one input image, feature description neural network (FDNN) is used to generate a new representation of this image, so that this image representation can be more efficiently compressed by standard codec, as compared to the input image. Furthermore, we use post-processing neural network (PPNN) to remove the coding artifacts caused by quantization of codec. Secondly, low-resolution image representation is adopted for high efficiency compression in terms of most of bit spent by image's structures under low bit-rate. However, more bits should be assigned to image details in the high-resolution, when most of structures have been kept after compression at the high bit-rate. This comes from a fact that the low-resolution image representation can't burden more information than high-resolution representation beyond a certain bit-rate. Finally, to resolve the problem of error back-propagation from the PPNN network to the FDNN network, we introduce to learn a virtual codec neural network to imitate two continuous procedures of standard compression and post-processing. The objective experimental results have demonstrated the proposed method has a large margin improvement, when comparing with several state-of-the-art approaches.

Lijun Zhao, Huihui Bai, Anhong Wang et al.

Multiple description coding (MDC) is able to stably transmit the signal in the un-reliable and non-prioritized networks, which has been broadly studied for several decades. However, the traditional MDC doesn't well leverage image's context features to generate multiple descriptions. In this paper, we propose a novel standard-compliant convolutional neural network-based MDC framework in term of image's context features. Firstly, multiple description generator network (MDGN) is designed to produce appearance-similar yet feature-different multiple descriptions automatically according to image's content, which are compressed by standard codec. Secondly, we present multiple description reconstruction network (MDRN) including side reconstruction network (SRN) and central reconstruction network (CRN). When any one of two lossy descriptions is received at the decoder, SRN network is used to improve the quality of this decoded lossy description by removing the compression artifact and up-sampling simultaneously. Meanwhile, we utilize CRN network with two decoded descriptions as inputs for better reconstruction, if both of lossy descriptions are available. Thirdly, multiple description virtual codec network (MDVCN) is proposed to bridge the gap between MDGN network and MDRN network in order to train an end-to-end MDC framework. Here, two learning algorithms are provided to train our whole framework. In addition to structural similarity loss function, the produced descriptions are used as opposing labels with multiple description distance loss function to regularize the training of MDGN network. These losses guarantee that the generated description images are structurally similar yet finely diverse. Experimental results show a great deal of objective and subjective quality measurements to validate the efficiency of the proposed method.

Lijun Zhao, Huihui Bai, Anhong Wang et al.

Although deep convolutional neural network has been proved to efficiently eliminate coding artifacts caused by the coarse quantization of traditional codec, it's difficult to train any neural network in front of the encoder for gradient's back-propagation. In this paper, we propose an end-to-end image compression framework based on convolutional neural network to resolve the problem of non-differentiability of the quantization function in the standard codec. First, the feature description neural network is used to get a valid description in the low-dimension space with respect to the ground-truth image so that the amount of image data is greatly reduced for storage or transmission. After image's valid description, standard image codec such as JPEG is leveraged to further compress image, which leads to image's great distortion and compression artifacts, especially blocking artifacts, detail missing, blurring, and ringing artifacts. Then, we use a post-processing neural network to remove these artifacts. Due to the challenge of directly learning a non-linear function for a standard codec based on convolutional neural network, we propose to learn a virtual codec neural network to approximate the projection from the valid description image to the post-processed compressed image, so that the gradient could be efficiently back-propagated from the post-processing neural network to the feature description neural network during training. Meanwhile, an advanced learning algorithm is proposed to train our deep neural networks for compression. Obviously, the priority of the proposed method is compatible with standard existing codecs and our learning strategy can be easily extended into these codecs based on convolutional neural network. Experimental results have demonstrated the advances of the proposed method as compared to several state-of-the-art approaches, especially at very low bit-rate.

Lijun Zhao, Huihui Bai, Jie Liang et al.

Recently, Generative Adversarial Network (GAN) has been found wide applications in style transfer, image-to-image translation and image super-resolution. In this paper, a color-depth conditional GAN is proposed to concurrently resolve the problems of depth super-resolution and color super-resolution in 3D videos. Firstly, given the low-resolution depth image and low-resolution color image, a generative network is proposed to leverage mutual information of color image and depth image to enhance each other in consideration of the geometry structural dependency of color-depth image in the same scene. Secondly, three loss functions, including data loss, total variation loss, and 8-connected gradient difference loss are introduced to train this generative network in order to keep generated images close to the real ones, in addition to the adversarial loss. Experimental results demonstrate that the proposed approach produces high-quality color image and depth image from low-quality image pair, and it is superior to several other leading methods. Besides, we use the same neural network framework to resolve the problem of image smoothing and edge detection at the same time.

Lijun Zhao, Jie Liang, Huihui Bai et al.

In this paper, two local activity-tuned filtering frameworks are proposed for noise removal and image smoothing, where the local activity measurement is given by the clipped and normalized local variance or standard deviation. The first framework is a modified anisotropic diffusion for noise removal of piece-wise smooth image. The second framework is a local activity-tuned Relative Total Variation (LAT-RTV) method for image smoothing. Both frameworks employ the division of gradient and the local activity measurement to achieve noise removal. In addition, to better capture local information, the proposed LAT-RTV uses the product of gradient and local activity measurement to boost the performance of image smoothing. Experimental results are presented to demonstrate the efficiency of the proposed methods on various applications, including depth image filtering, clip-art compression artifact removal, image smoothing, and image denoising.