Vector-Quantization (VQ) based discrete generative models are widely used to learn powerful high-quality (HQ) priors for blind image restoration (BIR). In this paper, we diagnose the side-effects of discrete VQ process essential to VQ-based BIR methods: 1) confining the representation capacity of HQ codebook, 2) being error-prone for code index prediction on low-quality (LQ) images, and 3) under-valuing the importance of input LQ image. These motivate us to learn continuous feature representation of HQ codebook for better restoration performance than using discrete VQ process. To further improve the restoration fidelity, we propose a new Self-in-Cross-Attention (SinCA) module to augment the HQ codebook with the feature of input LQ image, and perform cross-attention between LQ feature and input-augmented codebook. By this way, our SinCA leverages the input LQ image to enhance the representation of codebook for restoration fidelity. Experiments on four typical VQ-based BIR methods demonstrate that, by replacing the VQ process with a transformer using our SinCA, they achieve better quantitative and qualitative performance on blind image super-resolution and blind face restoration. The code and pre-trained models are publicly released at https://github.com/lhy-85/SinCA.

3D scene segmentation is a fundamental but challenging task for 3D scene understanding in computer vision. However, the majority of existing methods rely heavily on labor-intensive and costly human-annotated 2D or 3D labels. To address this issue, we propose a novel Object-Centric Feature Field for 3d scene segmentation, called OFF3D. Given the machine-generated inconsistent 2D masks, OFF3D can build an implicit object feature field to achieve consistent segmentation for all objects in the 3D scene. Specifically, OFF3D utilizes the continuous object features to represent object properties of each 3D point and proposes a Pseudo-Segment Clustering module to roughly locate objects in the 3D scene at the segment level and design a Confidence-Weighted Contrastive loss to achieve precise object segmentation by pixel-level feature optimization. Extensive experiments demonstrate the effectiveness of our method compared to state-of-the-art methods both quantitatively and qualitatively.

Camouflaged object detection (COD) and salient object detection (SOD) are two distinct yet closely-related computer vision tasks widely studied during the past decades. Though sharing the same purpose of segmenting an image into binary foreground and background regions, their distinction lies in the fact that COD focuses on concealed objects hidden in the image, while SOD concentrates on the most prominent objects in the image. Building universal segmentation models is currently a hot topic in the community. Previous works achieved good performance on certain task by stacking various hand-designed modules and multi-scale features. However, these careful task-specific designs also make them lose their potential as general-purpose architectures. Therefore, we hope to build general architectures that can be applied to both tasks. In this work, we propose a simple yet effective network (SENet) based on vision Transformer (ViT), by employing a simple design of an asymmetric ViT-based encoder-decoder structure, we yield competitive results on both tasks, exhibiting greater versatility than meticulously crafted ones. To enhance the performance of universal architectures on both tasks, we propose some general methods targeting some common difficulties of the two tasks. First, we use image reconstruction as an auxiliary task during training to increase the difficulty of training, forcing the network to have a better perception of the image as a whole to help with segmentation tasks. In addition, we propose a local information capture module (LICM) to make up for the limitations of the patch-level attention mechanism in pixel-level COD and SOD tasks and a dynamic weighted loss (DW loss) to solve the problem that small target samples are more difficult to locate and segment in both tasks. Finally, we also conduct a preliminary exploration of joint training, trying to use one model to complete two tasks simultaneously. Extensive experiments on multiple benchmark datasets demonstrate the effectiveness of our method. The code is available at https://github.com/linuxsino/SENet.

Reducing the radiation dose in CT scanning is important to alleviate the damage to the human health in clinical scenes. A promising way is to replace the normal-dose CT (NDCT) imaging by low-dose CT (LDCT) imaging with lower tube voltage and tube current. This often brings severe noise to the LDCT images, which adversely affects the diagnosis accuracy. Most of existing LDCT image denoising networks are trained either with synthetic LDCT images or real-world LDCT and NDCT image pairs with huge spatial misalignment. However, the synthetic noise is very different from the complex noise in real-world LDCT images, while the huge spatial misalignment brings inaccurate predictions of tissue structures in the denoised LDCT images. To well utilize real-world LDCT and NDCT image pairs for LDCT image denoising, in this paper, we introduce a new Patch Similarity Purification (PSP) strategy to construct high-quality training dataset for network training. Specifically, our PSP strategy first perform binarization for each pair of image patches cropped from the corresponding LDCT and NDCT image pairs. For each pair of binary masks, it then computes their similarity ratio by common mask calculation, and the patch pair can be selected as a training sample if their mask similarity ratio is higher than a threshold. By using our PSP strategy, each training set of our Rabbit and Patient datasets contain hundreds of thousands of real-world LDCT and NDCT image patch pairs with negligible misalignment. Extensive experiments demonstrate the usefulness of our PSP strategy on purifying the training data and the effectiveness of training LDCT image denoising networks on our datasets. The code and dataset are provided at https://github.com/TuTusong/PSP.

Accurate tissue segmentation of infant brain in magnetic resonance (MR) images is crucial for charting early brain development and identifying biomarkers. Due to ongoing myelination and maturation, in the isointense phase (6-9 months of age), the gray and white matters of infant brain exhibit similar intensity levels in MR images, posing significant challenges for tissue segmentation. Meanwhile, in the adult-like phase around 12 months of age, the MR images show high tissue contrast and can be easily segmented. In this paper, we propose to effectively exploit adult-like phase images to achieve robust multi-view isointense infant brain segmentation. Specifically, in one way, we transfer adult-like phase images to the isointense view, which have similar tissue contrast as the isointense phase images, and use the transferred images to train an isointense-view segmentation network. On the other way, we transfer isointense phase images to the adult-like view, which have enhanced tissue contrast, for training a segmentation network in the adult-like view. The segmentation networks of different views form a multi-path architecture that performs multi-view learning to further boost the segmentation performance. Since anatomy-preserving style transfer is key to the downstream segmentation task, we develop a Disentangled Cycle-consistent Adversarial Network (DCAN) with strong regularization terms to accurately transfer realistic tissue contrast between isointense and adult-like phase images while still maintaining their structural consistency. Experiments on both NDAR and iSeg-2019 datasets demonstrate a significant superior performance of our method over the state-of-the-art methods.

Metal dental implants may introduce metal artifacts (MA) during the CBCT imaging process, causing significant interference in subsequent diagnosis. In recent years, many deep learning methods for metal artifact reduction (MAR) have been proposed. Due to the huge difference between synthetic and clinical MA, supervised learning MAR methods may perform poorly in clinical settings. Many existing unsupervised MAR methods trained on clinical data often suffer from incorrect dental morphology. To alleviate the above problems, in this paper, we propose a new MAR method of Coupled Diffusion Models (CDM) for clinical dental CBCT images. Specifically, we separately train two diffusion models on clinical MA-degraded images and clinical clean images to obtain prior information, respectively. During the denoising process, the variances of noise levels are calculated from MA images and the prior of diffusion models. Then we develop a noise transformation module between the two diffusion models to transform the MA noise image into a new initial value for the denoising process. Our designs effectively exploit the inherent transformation between the misaligned MA-degraded images and clean images. Additionally, we introduce an MA-adaptive inference technique to better accommodate the MA degradation in different areas of an MA-degraded image. Experiments on our clinical dataset demonstrate that our CDM outperforms the comparison methods on both objective metrics and visual quality, especially for severe MA degradation. We will publicly release our code.

Fine-tuning pretrained large models to downstream tasks is an important problem, which however suffers from huge memory overhead due to large-scale parameters. This work strives to reduce memory overhead in fine-tuning from perspectives of activation function and layer normalization. To this end, we propose the Approximate Backpropagation (Approx-BP) theory, which provides the theoretical feasibility of decoupling the forward and backward passes. We apply our Approx-BP theory to backpropagation training and derive memory-efficient alternatives of GELU and SiLU activation functions, which use derivative functions of ReLUs in the backward pass while keeping their forward pass unchanged. In addition, we introduce a Memory-Sharing Backpropagation strategy, which enables the activation memory to be shared by two adjacent layers, thereby removing activation memory usage redundancy. Our method neither induces extra computation nor reduces training efficiency. We conduct extensive experiments with pretrained vision and language models, and the results demonstrate that our proposal can reduce up to $\sim$$30\%$ of the peak memory usage. Our code is released at [github](https://github.com/yyyyychen/LowMemoryBP).

The Space-Time Video Super-Resolution (STVSR) task aims to enhance the visual quality of videos, by simultaneously performing video frame interpolation (VFI) and video super-resolution (VSR). However, facing the challenge of the additional temporal dimension and scale inconsistency, most existing STVSR methods are complex and inflexible in dynamically modeling different motion amplitudes. In this work, we find that choosing an appropriate processing scale achieves remarkable benefits in flow-based feature propagation. We propose a novel Scale-Adaptive Feature Aggregation (SAFA) network that adaptively selects sub-networks with different processing scales for individual samples. Experiments on four public STVSR benchmarks demonstrate that SAFA achieves state-of-the-art performance. Our SAFA network outperforms recent state-of-the-art methods such as TMNet [83] and VideoINR [10] by an average improvement of over 0.5dB on PSNR, while requiring less than half the number of parameters and only 1/3 computational costs.

The illumination of improperly exposed photographs has been widely corrected using deep convolutional neural networks or Transformers. Despite with promising performance, these methods usually suffer from large parameter amounts and heavy computational FLOPs on high-resolution photographs. In this paper, we propose extremely light-weight (with only ~8K parameters) Multi-Scale Linear Transformation (MSLT) networks under the multi-layer perception architecture, which can process 4K-resolution sRGB images at 125 Frame-Per-Second (FPS) by a Titan RTX GPU. Specifically, the proposed MSLT networks first decompose an input image into high and low frequency layers by Laplacian pyramid techniques, and then sequentially correct different layers by pixel-adaptive linear transformation, which is implemented by efficient bilateral grid learning or 1 × 1 convolutions. Experiments on two benchmark datasets demonstrate the efficiency of our MSLTs against the state-of-the-arts on photo exposure correction. Extensive ablation studies validate the effectiveness of our contributions. The code is available at https://github.com/Zhou-Yijie/MSLTNet.

Few-shot learning deals with the fundamental and challenging problem of learning from a few annotated samples, while being able to generalize well on new tasks. The crux of few-shot learning is to extract prior knowledge from related tasks to enable fast adaptation to a new task with a limited amount of data. In this paper, we propose meta-learning kernels with random Fourier features for few-shot learning, we call MetaKernel. Specifically, we propose learning variational random features in a data-driven manner to obtain task-specific kernels by leveraging the shared knowledge provided by related tasks in a meta-learning setting. We treat the random feature basis as the latent variable, which is estimated by variational inference. The shared knowledge from related tasks is incorporated into a context inference of the posterior, which we achieve via a long-short term memory module. To establish more expressive kernels, we deploy conditional normalizing flows based on coupling layers to achieve a richer posterior distribution over random Fourier bases. The resultant kernels are more informative and discriminative, which further improves the few-shot learning. To evaluate our method, we conduct extensive experiments on both few-shot image classification and regression tasks. A thorough ablation study demonstrates that the effectiveness of each introduced component in our method. The benchmark results on fourteen datasets demonstrate MetaKernel consistently delivers at least comparable and often better performance than state-of-the-art alternatives.

Appearance-based gaze estimation has been widely studied recently with promising performance. The majority of appearance-based gaze estimation methods are developed under the deterministic frameworks. However, the deterministic gaze estimation methods suffer from large performance drop upon challenging eye images in low-resolution, darkness, partial occlusions, etc. To alleviate this problem, in this article, we alternatively reformulate the appearance-based gaze estimation problem under a generative framework. Specifically, we propose a variational inference model, that is, variational gaze estimation network (VGE-Net), to generate multiple gaze maps as complimentary candidates simultaneously supervised by the ground-truth gaze map. To achieve robust estimation, we adaptively fuse the gaze directions predicted on these candidate gaze maps by a regression network through a simple attention mechanism. Experiments on three benchmarks, that is, MPIIGaze, EYEDIAP, and Columbia, demonstrate that our VGE-Net outperforms state-of-the-art gaze estimation methods, especially on challenging cases. Comprehensive ablation studies also validate the effectiveness of our contributions. The code will be publicly released.

Synthesis of unavailable imaging modalities from available ones can generate modality-specific complementary information and enable multi-modality based medical images diagnosis or treatment. Existing generative methods for medical image synthesis are usually based on cross-modal translation between acquired and missing modalities. These methods are usually dedicated to specific missing modality and perform synthesis in one shot, which cannot deal with varying number of missing modalities flexibly and construct the mapping across modalities effectively. To address the above issues, in this paper, we propose a unified Multi-modal Modality-masked Diffusion Network (M2DN), tackling multi-modal synthesis from the perspective of "progressive whole-modality inpainting", instead of "cross-modal translation". Specifically, our M2DN considers the missing modalities as random noise and takes all the modalities as a unity in each reverse diffusion step. The proposed joint synthesis scheme performs synthesis for the missing modalities and self-reconstruction for the available ones, which not only enables synthesis for arbitrary missing scenarios, but also facilitates the construction of common latent space and enhances the model representation ability. Besides, we introduce a modality-mask scheme to encode availability status of each incoming modality explicitly in a binary mask, which is adopted as condition for the diffusion model to further enhance the synthesis performance of our M2DN for arbitrary missing scenarios. We carry out experiments on two public brain MRI datasets for synthesis and downstream segmentation tasks. Experimental results demonstrate that our M2DN outperforms the state-of-the-art models significantly and shows great generalizability for arbitrary missing modalities.

Few-shot image classification aims to recognize unseen classes with few labeled samples. Existing meta-learning models learn the ability of learning good representation or model parameters, in order to adapt to new tasks with a few training samples. However, when there exists a domain gap between training and test tasks, the learned ability often does not generalize well across domains, resulting in degraded performance on new tasks. In this article, we propose variational neuron shifting to generate adapted feature representations for few-shot learning. To do so, we introduce a working memory module to store the shifted neurons from the support set, which will be accessed to generate adapted feature representations of query samples. Under the meta-learning paradigm, the model is learned to acquire the ability of adaptation with single sample at meta-training time so as to further adapt itself to each single test sample at meta-test time. We formulate the adaptation process as a variational Bayesian inference problem, which incorporates the test sample as the condition into the generation of the model neuron shifting. We conduct extensive experiments on both within and across domain few-shot classification tasks. The new state-of-the-art performance substantiates the effectiveness of our variational neuron shifting. The thorough ablation studies further demonstrate the benefit of each component in our model.

Exploiting long-range semantic contexts and geometric information is crucial to infer salient objects from RGB and depth features. However, existing methods mainly focus on excavating local features within fixed regions by continuously feeding forward networks. In this article, we introduce Dynamic Message Propagation (DMP) to dynamically learn context information within more flexible regions. We integrate DMP into a Siamese-based network to process the RGB image and depth map separately and design a multi-level feature fusion module to explore cross-level information between refined RGB and depth features. Extensive experiments show clear improvements of our method over 17 methods on six benchmark datasets for RGB-D salient object detection (SOD). Additionally, our method outperforms its competitors for the video SOD task. Code is available at https://github.com/chenbaian-cs/DMPNet .

The performance of video prediction has been greatly boosted by advanced deep neural networks. However, most of the current methods suffer from large model sizes and require extra inputs, e.g., semantic/depth maps, for promising performance. For efficiency consideration, in this paper, we propose a Dynamic Multi-scale Voxel Flow Network (DMVFN) to achieve better video prediction performance at lower computational costs with only RGB images, than previous methods. The core of our DMVFN is a differentiable routing module that can effectively perceive the motion scales of video frames. Once trained, our DMVFN selects adaptive sub-networks for different inputs at the inference stage. Experiments on several benchmarks demonstrate that our DMVFN is an order of magnitude faster than Deep Voxel Flow [35] and surpasses the state-of-the-art iterative-based OPT [63] on generated image quality.

Real-world face super-resolution (SR) is a highly ill-posed image restoration task. The fully-cycled Cycle-GAN architecture is widely employed to achieve promising performance on face SR, but is prone to produce artifacts upon challenging cases in real-world scenarios, since joint participation in the same degradation branch will impact final performance due to huge domain gap between real-world and synthetic LR ones obtained by generators. To better exploit the powerful generative capability of GAN for real-world face SR, in this paper, we establish two independent degradation branches in the forward and backward cycle-consistent reconstruction processes, respectively, while the two processes share the same restoration branch. Our Semi-Cycled Generative Adversarial Networks (SCGAN) is able to alleviate the adverse effects of the domain gap between the real-world LR face images and the synthetic LR ones, and to achieve accurate and robust face SR performance by the shared restoration branch regularized by both the forward and backward cycle-consistent learning processes. Experiments on two synthetic and two real-world datasets demonstrate that, our SCGAN outperforms the state-of-the-art methods on recovering the face structures/details and quantitative metrics for real-world face SR. The code will be publicly released at https://github.com/HaoHou-98/SCGAN.

Unsupervised cross-domain crowd counting has recently received great attention in computer vision, which generalizes the model from the source domain to the unlabeled target domain. However, it is an extremely challenging task because only unlabeled data is available from the target domain and the domain gap between two domains is implicit in crowd counting. In this paper, we propose a latent domain generation method to improve the generalization ability of unsupervised domain adaptation crowd counting by generating a latent domain. To this end, we propose a domain generator with random perturbations to learn a new latent distribution derived from the original source distribution. The latent domain generator can extract target information sampled in its stochastic latent representation, which preserves the original target information and enhances the variational ability. Meanwhile, to ensure that the generated latent domain is consistent with the source domain in counting performance, we introduce a consistency loss to encourage similar output from latent and source domains. Moreover, to enhance the adaptation ability of the generated latent domain, we apply the adversarial loss to achieve alignment between the latent and target domains. The domain generator with the adversarial loss and consistency loss ensures that the generated domain is aligned to the target while also improving the robustness of the original source domain model. The experiment indicates that our framework can effortlessly extend to scenarios with different objects (crowd, cars). The experiments also demonstrate the effectiveness of our method on unsupervised realistic-to-realistic crowd counting problems.

Due to the constraints of the imaging device and high cost in operation time, computer tomography (CT) scans are usually acquired with low within-slice resolution. Improving the inter-slice resolution is beneficial to the disease diagnosis for both human experts and computer-aided systems. To this end, this paper builds a novel medical slice synthesis to increase the inter-slice resolution. Considering that the groundtruth intermediate medical slices are always absent in clinical practice, we introduce the incremental cross-view mutual distillation strategy to accomplish this task in the self-supervised learning manner. Specifically, we model this problem from three different views: slice-wise interpolation from axial view and pixel-wise interpolation from coronal and sagittal views. Under this circumstance, the models learned from different views can distill valuable knowledge to guide the learning processes of each other. We can repeat this process to make the models synthesize intermediate slice data with increasing between-slice resolution. To demonstrate the effectiveness of the proposed approach, we conduct comprehensive experiments on a large-scale$CT$dataset. Quantitative and qualitative comparison results show that our method outperforms state-of-the-art algorithms by clear margins.

The high computational cost of neural networks has prevented recent successes in RGB-D salient object detection (SOD) from benefiting real-world applications. Hence, this article introduces a novel network, MobileSal, which focuses on efficient RGB-D SOD using mobile networks for deep feature extraction. However, mobile networks are less powerful in feature representation than cumbersome networks. To this end, we observe that the depth information of color images can strengthen the feature representation related to SOD if leveraged properly. Therefore, we propose an implicit depth restoration (IDR) technique to strengthen the mobile networks' feature representation capability for RGB-D SOD. IDR is only adopted in the training phase and is omitted during testing, so it is computationally free. Besides, we propose compact pyramid refinement (CPR) for efficient multi-level feature aggregation to derive salient objects with clear boundaries. With IDR and CPR incorporated, MobileSal performs favorably against state-of-the-art methods on six challenging RGB-D SOD datasets with much faster speed (450fps for the input size of 320×320) and fewer parameters (6.5M). The code is released at https://mmcheng.net/mobilesal.

We focus on a fundamental task of detecting meaningful line structures, a.k.a., semantic line, in natural scenes. Many previous methods regard this problem as a special case of object detection and adjust existing object detectors for semantic line detection. However, these methods neglect the inherent characteristics of lines, leading to sub-optimal performance. Lines enjoy much simpler geometric property than complex objects and thus can be compactly parameterized by a few arguments. To better exploit the property of lines, in this paper, we incorporate the classical Hough transform technique into deeply learned representations and propose a one-shot end-to-end learning framework for line detection. By parameterizing lines with slopes and biases, we perform Hough transform to translate deep representations into the parametric domain, in which we perform line detection. Specifically, we aggregate features along candidate lines on the feature map plane and then assign the aggregated features to corresponding locations in the parametric domain. Consequently, the problem of detecting semantic lines in the spatial domain is transformed into spotting individual points in the parametric domain, making the post-processing steps, i.e., non-maximal suppression, more efficient. Furthermore, our method makes it easy to extract contextual line features that are critical for accurate line detection. In addition to the proposed method, we design an evaluation metric to assess the quality of line detection and construct a large scale dataset for the line detection task. Experimental results on our proposed dataset and another public dataset demonstrate the advantages of our method over previous state-of-the-art alternatives. The dataset and source code is available at https://mmcheng.net/dhtline/.

Unsupervised domain adaptation crowd counting (UDACC) has been studied with practical research utility by getting rid of the labeling burden on large-scale dense crowds in the target domain. Current methods generalize well within the specific domain gap by directly aligning domain distributions or translating synthetic data to realistic images. However, it is difficult to define domain gaps among complex real-world datasets, in which the images vary greatly in style, density level and/or content. To tackle this problem, in this paper, we propose a Cross-Domain Attention Network (CDANet), which can effectively generalize the model to the unlabeled domain on both unsupervised synthetic-to-realistic and realistic-to-realistic crowd counting. Specifically, we propose a Cross-Domain Attention Module (CDAM) to learn domain-related information between the source and target domain, which extracts relations in cross-domain attentive information, thus enhancing crowd-informative features. Moreover, to make our CDAM invariant to domain shifts, we introduce a consistency penalty to ensure that the attention maps are consistent before and after the domain shifting. Thus our CDANet can pay attention to the shared counting information across domains, while remaining its invariant ability during domain adaptation. Extensive experiments on several common benchmarks for UDACC demonstrate that our CDANet gets competitive results on both unsupervised synthetic-to-realistic and realistic-to-realistic UDACC tasks.

Space-time video super-resolution (STVSR) aims to increase the spatial and temporal resolutions of low-resolution and low-frame-rate videos. Recently, deformable convolution based methods have achieved promising STVSR performance, but they could only infer the intermediate frame pre-defined in the training stage. Besides, these methods undervalued the short-term motion cues among adjacent frames. In this paper, we propose a Temporal Modulation Network (TMNet) to interpolate arbitrary intermediate frame(s) with accurate high-resolution reconstruction. Specifically, we propose a Temporal Modulation Block (TMB) to modulate deformable convolution kernels for controllable feature interpolation. To well exploit the temporal information, we propose a Locally-temporal Feature Comparison (LFC) module, along with the Bi-directional Deformable ConvLSTM, to extract short-term and long-term motion cues in videos. Experiments on three benchmark datasets demonstrate that our TMNet outperforms previous STVSR methods. The code is available at https://github.com/CS-GangXu/TMNet.

The self-expressive property of data points, that is, each data point can be linearly represented by the other data points in the same subspace, has proven effective in leading subspace clustering (SC) methods. Most self-expressive methods usually construct a feasible affinity matrix from a coefficient matrix, obtained by solving an optimization problem. However, the negative entries in the coefficient matrix are forced to be positive when constructing the affinity matrix via exponentiation, absolute symmetrization, or squaring operations. This consequently damages the inherent correlations among the data. Besides, the affine constraint used in these methods is not flexible enough for practical applications. To overcome these problems, in this article, we introduce a scaled simplex representation (SSR) for the SC problem. Specifically, the non-negative constraint is used to make the coefficient matrix physically meaningful, and the coefficient vector is constrained to be summed up to a scalar to make it more discriminative. The proposed SSR-based SC (SSRSC) model is reformulated as a linear equality-constrained problem, which is solved efficiently under the alternating direction method of multipliers framework. Experiments on benchmark datasets demonstrate that the proposed SSRSC algorithm is very efficient and outperforms the state-of-the-art SC methods on accuracy. The code can be found at https://github.com/csjunxu/SSRSC.

Current RGB-D salient object detection (SOD) methods utilize the depth stream as complementary information to the RGB stream. However, the depth maps are usually of low-quality in existing RGB-D SOD datasets. Most RGB-D SOD networks trained with these datasets would produce error-prone results. In this paper, we propose a novel Complementary Depth Network (CDNet) to well exploit saliency-informative depth features for RGB-D SOD. To alleviate the influence of low-quality depth maps to RGB-D SOD, we propose to select saliency-informative depth maps as the training targets and leverage RGB features to estimate meaningful depth maps. Besides, to learn robust depth features for accurate prediction, we propose a new dynamic scheme to fuse the depth features extracted from the original and estimated depth maps with adaptive weights. What's more, we design a two-stage cross-modal feature fusion scheme to well integrate the depth features with the RGB ones, further improving the performance of our CDNet on RGB-D SOD. Experiments on seven benchmark datasets demonstrate that our CDNet outperforms state-of-the-art RGB-D SOD methods. The code is publicly available at https://github.com/blanclist/CDNet.

Recently, the coronavirus disease 2019 (COVID-19) has caused a pandemic disease in over 200 countries, influencing billions of humans. To control the infection, identifying and separating the infected people is the most crucial step. The main diagnostic tool is the Reverse Transcription Polymerase Chain Reaction (RT-PCR) test. Still, the sensitivity of the RT-PCR test is not high enough to effectively prevent the pandemic. The chest CT scan test provides a valuable complementary tool to the RT-PCR test, and it can identify the patients in the early-stage with high sensitivity. However, the chest CT scan test is usually time-consuming, requiring about 21.5 minutes per case. This paper develops a novel Joint Classification and Segmentation (JCS) system to perform real-time and explainable COVID- 19 chest CT diagnosis. To train our JCS system, we construct a large scale COVID- 19 Classification and Segmentation (COVID-CS) dataset, with 144,167 chest CT images of 400 COVID- 19 patients and 350 uninfected cases. 3,855 chest CT images of 200 patients are annotated with fine-grained pixel-level labels of opacifications, which are increased attenuation of the lung parenchyma. We also have annotated lesion counts, opacification areas, and locations and thus benefit various diagnosis aspects. Extensive experiments demonstrate that the proposed JCS diagnosis system is very efficient for COVID-19 classification and segmentation. It obtains an average sensitivity of 95.0% and a specificity of 93.0% on the classification test set, and 78.5% Dice score on the segmentation test set of our COVID-CS dataset. The COVID-CS dataset and code are available at https://github.com/yuhuan-wu/JCS.

RGB-D salient object detection (SOD) aims to segment the most attractive objects in a pair of cross-modal RGB and depth images. Currently, most existing RGB-D SOD methods focus on the foreground region when utilizing the depth images. However, the background also provides important information in traditional SOD methods for promising performance. To better explore salient information in both foreground and background regions, this paper proposes a Bilateral Attention Network (BiANet) for the RGB-D SOD task. Specifically, we introduce a Bilateral Attention Module (BAM) with a complementary attention mechanism: foreground-first (FF) attention and background-first (BF) attention. The FF attention focuses on the foreground region with a gradual refinement style, while the BF one recovers potentially useful salient information in the background region. Benefited from the proposed BAM module, our BiANet can capture more meaningful foreground and background cues, and shift more attention to refining the uncertain details between foreground and background regions. Additionally, we extend our BAM by leveraging the multi-scale techniques for better SOD performance. Extensive experiments on six benchmark datasets demonstrate that our BiANet outperforms other state-of-the-art RGB-D SOD methods in terms of objective metrics and subjective visual comparison. Our BiANet can run up to 80 fps on 224×224 RGB-D images, with an NVIDIA GeForce RTX 2080Ti GPU. Comprehensive ablation studies also validate our contributions.

Recently, imagesmoothing has gained increasing attention due to its prerequisite role in other image processing tasks, e.g., image enhancement and editing. However, the evaluation of image smoothing algorithms is usually performed by subjective observation on images without corresponding ground truths. To promote the development of image smoothing algorithms, in this paper, we construct a novel Nankai Smoothing (NKS) dataset containing 200 images blended by versatile structure images and natural textures. The structure images are inherently smooth and naturally taken as ground truths. On our NKS dataset, we comprehensively evaluate 14 popular image smoothing algorithms. Moreover, we propose a Pixel-level Non-Local Smoothing (PNLS) method to well preserve the structure of the smoothed images, by exploiting the pixel-level non-local self-similarity prior of natural images. Extensive experiments on several benchmark datasets demonstrate that our PNLS outperforms previous algorithms on the image smoothing task. Ablation studies also reveal the work mechanism of our PNLS on image smoothing. To further show its effectiveness, we apply our PNLS on several applications such as semantic region smoothing, detail/edge enhancement, and image abstraction. The dataset and code are available athttps://github.com/zal0302/PNLS.

We introduce kernels with random Fourier features in the meta-learning framework for few-shot learning. We propose meta variational random features (MetaVRF) to learn adaptive kernels for the base-learner, which is developed in a latent variable model by treating the random feature basis as the latent variable. We formulate the optimization of MetaVRF as a variational inference problem by deriving an evidence lower bound under the meta-learning framework. To incorporate shared knowledge from related tasks, we propose a context inference of the posterior, which is established by an LSTM architecture. The LSTM-based inference network can effectively integrate the context information of previous tasks with task-specific information, generating informative and adaptive features. The learned MetaVRF can produce kernels of high representational power with a relatively low spectral sampling rate and also enables fast adaptation to new tasks. Experimental results on a variety of few-shot regression and classification tasks demonstrate that MetaVRF delivers much better, or at least competitive, performance compared to existing meta-learning alternatives.

Intra-saliency and inter-saliency cues have been extensively studied for co-saliency detection (Co-SOD). Model-based methods produce coarse Co-SOD results due to hand-crafted intra- and inter-saliency features. Current data-driven models exploit inter-saliency cues, but undervalue the potential power of intra-saliency cues. In this paper, we propose an Intra-saliency Correlation Network (ICNet) to extract intra-saliency cues from the single image saliency maps (SISMs) predicted by any off-the-shelf SOD method, and obtain inter-saliency cues by correlation techniques. Specifically, we adopt normalized masked average pooling (NMAP) to extract latent intra-saliency categories from the SISMs and semantic features as intra cues. Then we employ a correlation fusion module (CFM) to obtain inter cues by exploiting correlations between the intra cues and single-image features. To improve Co-SOD performance, we propose a category-independent rearranged self-correlation feature (RSCF) strategy. Experiments on three benchmarks show that our ICNet outperforms previous state-of-the-art methods on Co-SOD. Ablation studies validate the effectiveness of our contributions. The PyTorch code is available at https://github.com/blanclist/ICNet.

Images captured in severe weather such as rain and snow significantly degrade the accuracy of vision systems, e.g., for outdoor video surveillance or autonomous driving. Image deraining is a critical yet highly challenging task, due to the fact that rain density varies across spatial locations, while the distribution patterns simultaneously vary across color channels. In this paper, we propose a variational image deraining (VID) method by formulating image deraining in a conditional variational auto-encoder framework. To achieve adaptive deraining to spatial rain density, we generate a density estimation map for each color channel, which can largely avoid over and under deraining. In addition, to address cross-channel variations, we conduct channel-wise deraining, motivated by our observation that bright pixels do not tend to remain bright after deraining unless their color channels are handled separately. Experimental results show that the proposed deraining method achieves superior performance on both synthesized and real rainy images, surpassing previous state-of-the-art methods by large margins.

Image deraining is an important yet challenging image processing task. Though deterministic image deraining methods are developed with encouraging performance, they are infeasible to learn flexible representations for probabilistic inference and diverse predictions. Besides, rain intensity varies both in spatial locations and across color channels, making this task more difficult. In this paper, we propose a Conditional Variational Image Deraining (CVID) network for better deraining performance, leveraging the exclusive generative ability of Conditional Variational Auto-Encoder (CVAE) on providing diverse predictions for the rainy image. To perform spatially adaptive deraining, we propose a spatial density estimation (SDE) module to estimate a rain density map for each image. Since rain density varies across different color channels, we also propose a channel-wise (CW) deraining scheme. Experiments on synthesized and real-world datasets show that the proposed CVID network achieves much better performance than previous deterministic methods on image deraining. Extensive ablation studies validate the effectiveness of the proposed SDE module and CW scheme in our CVID network. The code is available at https://github.com/Yingjun-Du/VID.

Non-local self similarity (NSS) is a powerful prior of natural images for image denoising. Most of existing denoising methods employ similar patches, which is a patch-level NSS prior. In this paper, we take one step forward by introducing a pixel-level NSS prior, i.e., searching similar pixels across a non-local region. This is motivated by the fact that finding closely similar pixels is more feasible than similar patches in natural images, which can be used to enhance image denoising performance. With the introduced pixel-level NSS prior, we propose an accurate noise level estimation method, and then develop a blind image denoising method based on the lifting Haar transform and Wiener filtering techniques. Experiments on benchmark datasets demonstrate that, the proposed method achieves much better performance than previous non-deep methods, and is still competitive with existing state-of-the-art deep learning based methods on real-world image denoising. The code is publicly available athttps://github.com/njusthyk1972/NLH.

Supervised deep networks have achieved promising performance on image denoising, by learning image priors and noise statistics on plenty pairs of noisy and clean images. Unsupervised denoising networks are trained with only noisy images. However, for an unseen corrupted image, both supervised and unsupervised networks ignore either its particular image prior, the noise statistics, or both. That is, the networks learned from external images inherently suffer from a domain gap problem: the image priors and noise statistics are very different between the training and test images. This problem becomes more clear when dealing with the signal dependent realistic noise. To circumvent this problem, in this work, we propose a novel "Noisy-As-Clean" (NAC) strategy of training self-supervised denoising networks. Specifically, the corrupted test image is directly taken as the "clean" target, while the inputs are synthetic images consisted of this corrupted image and a second yet similar corruption. A simple but useful observation on our NAC is: as long as the noise is weak, it is feasible to learn a self-supervised network only with the corrupted image, approximating the optimal parameters of a supervised network learned with pairs of noisy and clean images. Experiments on synthetic and realistic noise removal demonstrate that, the DnCNN and ResNet networks trained with our self-supervised NAC strategy achieve comparable or better performance than the original ones and previous supervised/unsupervised/self-supervised networks. The code is publicly available at https://github.com/csjunxu/Noisy-As-Clean.

Retinex theory is developed mainly to decompose an image into the illumination and reflectance components by analyzing local image derivatives. In this theory, larger derivatives are attributed to the changes in reflectance, while smaller derivatives are emerged in the smooth illumination. In this paper, we utilize exponentiated local derivatives (with an exponent γ) of an observed image to generate its structure map and texture map. The structure map is produced by been amplified with γ > 1, while the texture map is generated by been shrank with γ < 1. To this end, we design exponential filters for the local derivatives, and present their capability on extracting accurate structure and texture maps, influenced by the choices of exponents γ. The extracted structure and texture maps are employed to regularize the illumination and reflectance components in Retinex decomposition. A novel Structure and Texture Aware Retinex (STAR) model is further proposed for illumination and reflectance decomposition of a single image. We solve the STAR model by an alternating optimization algorithm. Each sub-problem is transformed into a vectorized least squares regression, with closed-form solutions. Comprehensive experiments on commonly tested datasets demonstrate that, the proposed STAR model produce better quantitative and qualitative performance than previous competing methods, on illumination and reflectance decomposition, low-light image enhancement, and color correction. The code is publicly available at https://github.com/csjunxu/STAR.

Deep neural networks (DNNs) are widely used and demonstrated their power in many applications, such as computer vision and pattern recognition. However, the training of these networks can be time consuming. Such a problem could be alleviated by using efficient optimizers. As one of the most commonly used optimizers, stochastic gradient descent-momentum (SGD-M) uses past and present gradients for parameter updates. However, in the process of network training, SGD-M may encounter some drawbacks, such as the overshoot phenomenon. This problem would slow the training convergence. To alleviate this problem and accelerate the convergence of DNN optimization, we propose a proportional-integral-derivative (PID) approach. Specifically, we investigate the intrinsic relationships between the PID-based controller and SGD-M first. We further propose a PID-based optimization algorithm to update the network parameters, where the past, current, and change of gradients are exploited. Consequently, our proposed PID-based optimization alleviates the overshoot problem suffered by SGD-M. When tested on popular DNN architectures, it also obtains up to 50% acceleration with competitive accuracy. Extensive experiments about computer vision and natural language processing demonstrate the effectiveness of our method on benchmark data sets, including CIFAR10, CIFAR100, Tiny-ImageNet, and PTB. We have released the code at https://github.com/tensorboy/PIDOptimizer.

Despite online learning (OL) techniques have boosted the performance of semi-supervised video object segmentation (VOS) methods, the huge time costs of OL greatly restricts their practicality. Matching based and propagation based methods run at a faster speed by avoiding OL techniques. However, they are limited by sub-optimal accuracy, due to mismatching and drifting problems. In this paper, we develop a real-time yet very accurate Ranking Attention Network (RANet) for VOS. Specifically, to integrate the insights of matching based and propagation based methods, we employ an encoder-decoder framework to learn pixel-level similarity and segmentation in an end-to-end manner. To better utilize the similarity maps, we propose a novel ranking attention module, which automatically ranks and selects these maps for fine-grained VOS performance. Experiments on DAVIS16 and DAVIS17 datasets show that our RANet achieves the best speed-accuracy trade-off, e.g., with 33 milliseconds per frame and J&F=85.5% on DAVIS16. With OL, our RANet reaches J&F=87.1% on DAVIS16, exceeding state-of-the-art VOS methods. The code can be found at https://github.com/Storife/RANet.

Cross Version Defect Prediction (CVDP) is a practical scenario by training the classification model on the historical data of the prior version and then predicting the defect labels of modules in the current version. Unfortunately, the differences of data distribution across versions may hinder the effectiveness of the trained CVDP model. Thus, it is not trivial to select a suitable training subset from the prior version to promote the CVDP performance. In this paper, we propose a novel method, called Two-Stage Training Subset Selection (TSTSS), to address this challenging issue. In the first stage, TSTSS utilizes a sparse modeling representative selection method to select an initial module subset from the prior version which can well reconstruct the data of the prior version. In the second stage, TSTSS leverages a dissimilarity-based sparse subset selection method to further refine the selected module subset, which enables the selected modules to well represent the modules of the current version. Finally, we use a novel weighted extreme learning machine classifier to construct the CVDP model. We evaluate the CVDP performance of TSTSS on 50 cross-version pairs using 6 indicators. The experiments show that TSTSS can efficiently improve the CVDP performance compared with 11 baseline methods.

Deep neural networks have demonstrated their power in many computer vision applications. State-of-the-art deep architectures such as VGG, ResNet, and DenseNet are mostly optimized by the SGD-Momentum algorithm, which updates the weights by considering their past and current gradients. Nonetheless, SGD-Momentum suffers from the overshoot problem, which hinders the convergence of network training. Inspired by the prominent success of proportional-integral-derivative (PID) controller in automatic control, we propose a PID approach for accelerating deep network optimization. We first reveal the intrinsic connections between SGD-Momentum and PID based controller, then present the optimization algorithm which exploits the past, current, and change of gradients to update the network parameters. The proposed PID method reduces much the overshoot phenomena of SGD-Momentum, and it achieves up to 50% acceleration on popular deep network architectures with competitive accuracy, as verified by our experiments on the benchmark datasets including CIFAR10, CIFAR100, and Tiny-ImageNet.

Tone mapping aims to reproduce a standard dynamic range image from a high dynamic range image with visual information preserved. State-of-the-art tone mapping algorithms mostly decompose an image into a base layer and a detail layer, and process them accordingly. These methods may have problems of halo artifacts and over-enhancement, due to the lack of proper priors imposed on the two layers. In this paper, we propose a hybrid ℓ1-ℓ0decomposition model to address these problems. Specifically, an ℓ1sparsity term is imposed on the base layer to model its piecewise smoothness property. An ℓ0sparsity term is imposed on the detail layer as a structural prior, which leads to piecewise constant effect. We further propose a multiscale tone mapping scheme based on our layer decomposition model. Experiments show that our tone mapping algorithm achieves visually compelling results with little halo artifacts, outperforming the state-of-the-art tone mapping algorithms in both subjective and objective evaluations.

Software defect prediction aims at detecting the defect-prone software modules by mining historical development data from software repositories. If such modules are identified at the early stage of the development, it can save large amounts of resources. Cross Version Defect Prediction (CVDP) is a practical scenario by training the classification model on the historical data of the prior version and then predicting the defect labels of modules of the current version. However, software development is a constantly-evolving process which leads to the data distribution differences across versions within the same project. The distribution differences will degrade the performance of the classification model. In this paper, we approach this issue by leveraging a state-of-the-art Dissimilarity-based Sparse Subset Selection (DS3) method. This method selects a representative module subset from the prior version based on the pairwise dissimilarities between the modules of two versions and assigns each module of the current version to one of the representative modules. These selected modules can well represent the modules of the current version, thus mitigating the distribution differences. We evaluate the effectiveness of DS3 for CVDP performance on total 40 cross-version pairs from 56 versions of 15 projects with three traditional and two effort-aware indicators. The extensive experiments show that DS3 outperforms three baseline methods, especially in terms of two effort-aware indicators.

Most non-blind deconvolution methods are developed under the error-free kernel assumption, and are not robust to inaccurate blur kernel. Unfortunately, despite the great progress in blind deconvolution, estimation error remains inevitable during blur kernel estimation. Consequently, severe artifacts such as ringing effects and distortions are likely to be introduced in the non-blind deconvolution stage. In this paper, we tackle this issue by suggesting: 1) a partial map in the Fourier domain for modeling kernel estimation error, and 2) a partial deconvolution model for robust deblurring with inaccurate blur kernel. The partial map is constructed by detecting the reliable Fourier entries of estimated blur kernel. And partial deconvolution is applied to wavelet-based and learning-based models to suppress the adverse effect of kernel estimation error. Furthermore, an E-M algorithm is developed for estimating the partial map and recovering the latent sharp image alternatively. Experimental results show that our partial deconvolution model is effective in relieving artifacts caused by inaccurate blur kernel, and can achieve favorable deblurring quality on synthetic and real blurry images.

Most of existing image denoising methods learn image priors from either external data or the noisy image itself to remove noise. However, priors learned from external data may not be adaptive to the image to be denoised, while priors learned from the given noisy image may not be accurate due to the interference of corrupted noise. Meanwhile, the noise in real-world noisy images is very complex, which is hard to be described by simple distributions such as Gaussian distribution, making real-world noisy image denoising a very challenging problem. We propose to exploit the information in both external data and the given noisy image, and develop an external prior guided internal prior learning method for real-world noisy image denoising. We first learn external priors from an independent set of clean natural images. With the aid of learned external priors, we then learn internal priors from the given noisy image to refine the prior model. The external and internal priors are formulated as a set of orthogonal dictionaries to efficiently reconstruct the desired image. Extensive experiments are performed on several real-world noisy image datasets. The proposed method demonstrates highly competitive denoising performance, outperforming state-of-the-art denoising methods including those designed for real-world noisy images.

Most of the existing denoising algorithms are developed for grayscale images. It is not trivial to extend them for color image denoising since the noise statistics in R, G, and B channels can be very different for real noisy images. In this paper, we propose a multi-channel (MC) optimization model for real color image denoising under the weighted nuclear norm minimization (WNNM) framework. We concatenate the RGB patches to make use of the channel redundancy, and introduce a weight matrix to balance the data fidelity of the three channels in consideration of their different noise statistics. The proposed MC-WNNM model does not have an analytical solution. We reformulate it into a linear equality-constrained problem and solve it via alternating direction method of multipliers. Each alternative updating step has a closed-form solution and the convergence can be guaranteed. Experiments on both synthetic and real noisy image datasets demonstrate the superiority of the proposed MC-WNNM over state-of-the-art denoising methods.

Patch based image modeling has achieved a great success in low level vision such as image denoising. In particular, the use of image nonlocal self-similarity (NSS) prior, which refers to the fact that a local patch often has many nonlocal similar patches to it across the image, has significantly enhanced the denoising performance. However, in most existing methods only the NSS of input degraded image is exploited, while how to utilize the NSS of clean natural images is still an open problem. In this paper, we propose a patch group (PG) based NSS prior learning scheme to learn explicit NSS models from natural images for high performance denoising. PGs are extracted from training images by putting nonlocal similar patches into groups, and a PG based Gaussian Mixture Model (PG-GMM) learning algorithm is developed to learn the NSS prior. We demonstrate that, owe to the learned PG-GMM, a simple weighted sparse coding model, which has a closed-form solution, can be used to perform image denoising effectively, resulting in high PSNR measure, fast speed, and particularly the best visual quality among all competing methods.