arXiv subjects

Xinggang Wang

Publications and source records attributed to Xinggang Wang.

At least 199 records · Page 11Linked to original sources

arXiv2021

Deep Online Correction for Monocular Visual Odometry

In this work, we propose a novel deep online correction (DOC) framework for monocular visual odometry. The whole pipeline has two stages: First, depth maps and initial poses are obtained from convolutional neural networks (CNNs) trained in self-supervised manners. Second, the poses predicted by CNNs are further improved by minimizing photometric errors via gradient updates of poses during inference phases. The benefits of our proposed method are twofold: 1) Different from online-learning methods, DOC does not need to calculate gradient propagation for parameters of CNNs. Thus, it saves more computation resources during inference phases. 2) Unlike hybrid methods that combine CNNs with traditional methods, DOC fully relies on deep learning (DL) frameworks. Though without complex back-end optimization modules, our method achieves outstanding performance with relative transform error (RTE) = 2.0% on KITTI Odometry benchmark for Seq. 09, which outperforms traditional monocular VO frameworks and is comparable to hybrid methods.

cs.CV↗

arXiv2021

AlignSeg: Feature-Aligned Segmentation Networks

Aggregating features in terms of different convolutional blocks or contextual embeddings has been proven to be an effective way to strengthen feature representations for semantic segmentation. However, most of the current popular network architectures tend to ignore the misalignment issues during the feature aggregation process caused by 1) step-by-step downsampling operations, and 2) indiscriminate contextual information fusion. In this paper, we explore the principles in addressing such feature misalignment issues and inventively propose Feature-Aligned Segmentation Networks (AlignSeg). AlignSeg consists of two primary modules, i.e., the Aligned Feature Aggregation (AlignFA) module and the Aligned Context Modeling (AlignCM) module. First, AlignFA adopts a simple learnable interpolation strategy to learn transformation offsets of pixels, which can effectively relieve the feature misalignment issue caused by multiresolution feature aggregation. Second, with the contextual embeddings in hand, AlignCM enables each pixel to choose private custom contextual information in an adaptive manner, making the contextual embeddings aligned better to provide appropriate guidance. We validate the effectiveness of our AlignSeg network with extensive experiments on Cityscapes and ADE20K, achieving new state-of-the-art mIoU scores of 82.6% and 45.95%, respectively. Our source code will be made available.

cs.CV↗

arXiv2021

Learning to Focus: Cascaded Feature Matching Network for Few-shot Image Recognition

Deep networks can learn to accurately recognize objects of a category by training on a large number of annotated images. However, a meta-learning challenge known as a low-shot image recognition task comes when only a few images with annotations are available for learning a recognition model for one category. The objects in testing/query and training/support images are likely to be different in size, location, style, and so on. Our method, called Cascaded Feature Matching Network (CFMN), is proposed to solve this problem. We train the meta-learner to learn a more fine-grained and adaptive deep distance metric by focusing more on the features that have high correlations between compared images by the feature matching block which can align associated features together and naturally ignore those non-discriminative features. By applying the proposed feature matching block in different layers of the few-shot recognition network, multi-scale information among the compared images can be incorporated into the final cascaded matching feature, which boosts the recognition performance further and generalizes better by learning on relationships. The experiments for few-shot learning on two standard datasets, \emph{mini}ImageNet and Omniglot, have confirmed the effectiveness of our method. Besides, the multi-label few-shot task is first studied on a new data split of COCO which further shows the superiority of the proposed feature matching network when performing few-shot learning in complex images. The code will be made publicly available.

cs.CV↗

arXiv2020

ResizeMix: Mixing Data with Preserved Object Information and True Labels

Data augmentation is a powerful technique to increase the diversity of data, which can effectively improve the generalization ability of neural networks in image recognition tasks. Recent data mixing based augmentation strategies have achieved great success. Especially, CutMix uses a simple but effective method to improve the classifiers by randomly cropping a patch from one image and pasting it on another image. To further promote the performance of CutMix, a series of works explore to use the saliency information of the image to guide the mixing. We systematically study the importance of the saliency information for mixing data, and find that the saliency information is not so necessary for promoting the augmentation performance. Furthermore, we find that the cutting based data mixing methods carry two problems of label misallocation and object information missing, which cannot be resolved simultaneously. We propose a more effective but very easily implemented method, namely ResizeMix. We mix the data by directly resizing the source image to a small patch and paste it on another image. The obtained patch preserves more substantial object information compared with conventional cut-based methods. ResizeMix shows evident advantages over CutMix and the saliency-guided methods on both image classification and object detection tasks without additional computation cost, which even outperforms most costly search-based automatic augmentation methods.

cs.CV↗

arXiv2020

FNA++: Fast Network Adaptation via Parameter Remapping and Architecture Search

Deep neural networks achieve remarkable performance in many computer vision tasks. Most state-of-the-art (SOTA) semantic segmentation and object detection approaches reuse neural network architectures designed for image classification as the backbone, commonly pre-trained on ImageNet. However, performance gains can be achieved by designing network architectures specifically for detection and segmentation, as shown by recent neural architecture search (NAS) research for detection and segmentation. One major challenge though is that ImageNet pre-training of the search space representation (a.k.a. super network) or the searched networks incurs huge computational cost. In this paper, we propose a Fast Network Adaptation (FNA++) method, which can adapt both the architecture and parameters of a seed network (e.g. an ImageNet pre-trained network) to become a network with different depths, widths, or kernel sizes via a parameter remapping technique, making it possible to use NAS for segmentation and detection tasks a lot more efficiently. In our experiments, we apply FNA++ on MobileNetV2 to obtain new networks for semantic segmentation, object detection, and human pose estimation that clearly outperform existing networks designed both manually and by NAS. We also implement FNA++ on ResNets and NAS networks, which demonstrates a great generalization ability. The total computation cost of FNA++ is significantly less than SOTA segmentation and detection NAS approaches: 1737x less than DPC, 6.8x less than Auto-DeepLab, and 8.0x less than DetNAS. A series of ablation studies are performed to demonstrate the effectiveness, and detailed analysis is provided for more insights into the working mechanism. Codes are available at https://github.com/JaminFong/FNA.

cs.CV↗

arXiv2020

IoU-balanced Loss Functions for Single-stage Object Detection

Single-stage object detectors have been widely applied in computer vision applications due to their high efficiency. However, we find that the loss functions adopted by single-stage object detectors hurt the localization accuracy seriously. Firstly, the standard cross-entropy loss for classification is independent of the localization task and drives all the positive examples to learn as high classification scores as possible regardless of localization accuracy during training. As a result, there will be many detections that have high classification scores but low IoU or detections that have low classification scores but high IoU. Secondly, for the standard smooth L1 loss, the gradient is dominated by the outliers that have poor localization accuracy during training. The above two problems will decrease the localization accuracy of single-stage detectors. In this work, IoU-balanced loss functions that consist of IoU-balanced classification loss and IoU-balanced localization loss are proposed to solve the above problems. The IoU-balanced classification loss pays more attention to positive examples with high IoU and can enhance the correlation between classification and localization tasks. The IoU-balanced localization loss decreases the gradient of examples with low IoU and increases the gradient of examples with high IoU, which can improve the localization accuracy of models. Extensive experiments on challenging public datasets such as MS COCO, PASCAL VOC and Cityscapes demonstrate that both IoU-balanced losses can bring substantial improvement for the popular single-stage detectors, especially for the localization accuracy. On COCO test-dev, the proposed methods can substantially improve AP by $1.0\%\sim1.7\%$ and AP75 by $1.0\%\sim2.4\%$. On PASCAL VOC, it can also substantially improve AP by $1.3\%\sim1.5\%$ and AP80, AP90 by $1.6\%\sim3.9\%$.

cs.CV↗

arXiv2020

EfficientPose: Efficient Human Pose Estimation with Neural Architecture Search

Human pose estimation from image and video is a vital task in many multimedia applications. Previous methods achieve great performance but rarely take efficiency into consideration, which makes it difficult to implement the networks on resource-constrained devices. Nowadays real-time multimedia applications call for more efficient models for better interactions. Moreover, most deep neural networks for pose estimation directly reuse the networks designed for image classification as the backbone, which are not yet optimized for the pose estimation task. In this paper, we propose an efficient framework targeted at human pose estimation including two parts, the efficient backbone and the efficient head. By implementing the differentiable neural architecture search method, we customize the backbone network design for pose estimation and reduce the computation cost with negligible accuracy degradation. For the efficient head, we slim the transposed convolutions and propose a spatial information correction module to promote the performance of the final prediction. In experiments, we evaluate our networks on the MPII and COCO datasets. Our smallest model has only 0.65 GFLOPs with 88.1% PCKh@0.5 on MPII and our large model has only 2 GFLOPs while its accuracy is competitive with the state-of-the-art large model, i.e., HRNet with 9.5 GFLOPs.

cs.CV↗

arXiv2020

Medial Injury/Dysfunction Induced Granulation Tissue Repair is the Pathogenesis of Atherosclerosis

Atherosclerosis, a chronic lesion of vascular wall, remains a leading cause of death and loss of life years. Classical hypotheses for atherosclerosis are long-standing mainly to explain atherogenesis. Unfortunately, these hypotheses may not explain the variation in the susceptibility to atherosclerosis. These issues are controversial over the past 150 years. Atherosclerosis from human coronary arteries was examined and triangle of media was found to be a true portraiture of cells injury in the media, and triangle of intima was a true portraiture of myofibroblast proliferation, extracellular matrix (ECM) secretion, collagen fiber formation and intimal thickening to repair media dysfunction. Myofibroblasts, ECM and lumen (intima)/vasa vasorum (VV) (adventitia) constitute granulation tissue repair. With granulation tissue hyperplasia, lots of collagen fibers (normal or denatured), foam cells and new capillaries formed. Thus, the following theory was postulated: Risk factors induce smooth muscle cells (SMCs) injury/loss, and fibrosis or structure destruction could be developed in the media, which lead to media dysfunction. Media dysfunction prompts disturbed mechanical properties of blood vessels, resulting in bigger pressure buildup in the intima and adventitia. Granulation tissues in the intima/adventitia develop to repair the injured media. Atherosclerosis, stiffening or aneurysm develops depending on media dysfunction severity and granulated tissue repair mode/degree. Nearly all characteristics of clinical atherosclerosis could be ideally interpreted with the theory. We believe that media dysfunction is a key initiator in the pathogenesis of atherosclerosis. It is expected that media dysfunction theory of atherosclerosis, should offer better understanding of the etiology for atherosclerosis.

q-bio.TO↗

arXiv2020

Hemodynamic Bigger Hydrostatic Pressure Instead of Lower Shear Stress Aggravates Atherosclerosis

The traditional view is that low shear stress of blood flow accelerates atherosclerosis. How does a low shear stress damage the vessel or aggravate atherosclerosis? It is a puzzling question. Furthermore, shear stress of veins is similar to that of Relatively Vulnerable Zones (RVZ) of arteries, while intimal thickening of veins is not obvious as the arteries. There are many places in arterioles with lower shear stress as the decrease of blood velocity, while the thickening of arterioles is not as obvious as that of the medium-sized arteries. These issues make it unconvincing for us to explain atherogenesis with lower shear stress. According to Bernoulli fluid equation (P+1/2\r{ho}v^2+\r{ho}gh=constant), in a closed pipe full of flowing blood, at the same height, the hydrostatic pressure is inversely proportional to the second power of blood velocity. At any point of per unit mass of fluid micro cluster here, the reduction of 1/2\r{ho}v^2 would be converted into P. Therefore, hydrostatic pressure(P) is inversely proportional to v^2 in a very short distance or in the same transection of the artery. When blood micro cluster flows over a very short distance or the same transection of the artery, previous studies did not consider the conversion between 1/2\r{ho}v^2 and P. Therefore, low shear stress aggravates atherosclerosis is an appearance, and the essence is that these areas with smaller blood velocity have much bigger hydrostatic pressure, which aggravates atherosclerosis. This perfectly explains the predisposing sites of atherosclerosis.

q-bio.TO↗

arXiv2020

Boundary-preserving Mask R-CNN

Tremendous efforts have been made to improve mask localization accuracy in instance segmentation. Modern instance segmentation methods relying on fully convolutional networks perform pixel-wise classification, which ignores object boundaries and shapes, leading coarse and indistinct mask prediction results and imprecise localization. To remedy these problems, we propose a conceptually simple yet effective Boundary-preserving Mask R-CNN (BMask R-CNN) to leverage object boundary information to improve mask localization accuracy. BMask R-CNN contains a boundary-preserving mask head in which object boundary and mask are mutually learned via feature fusion blocks. As a result, the predicted masks are better aligned with object boundaries. Without bells and whistles, BMask R-CNN outperforms Mask R-CNN by a considerable margin on the COCO dataset; in the Cityscapes dataset, there are more accurate boundary groundtruths available, so that BMask R-CNN obtains remarkable improvements over Mask R-CNN. Besides, it is not surprising to observe that BMask R-CNN obtains more obvious improvement when the evaluation criterion requires better localization (e.g., AP$_{75}$) as shown in Fig.1. Code and models are available at \url{https://github.com/hustvl/BMaskR-CNN}.

cs.CV↗

arXiv2020

Deep multi-metric learning for text-independent speaker verification

Text-independent speaker verification is an important artificial intelligence problem that has a wide spectrum of applications, such as criminal investigation, payment certification, and interest-based customer services. The purpose of text-independent speaker verification is to determine whether two given uncontrolled utterances originate from the same speaker or not. Extracting speech features for each speaker using deep neural networks is a promising direction to explore and a straightforward solution is to train the discriminative feature extraction network by using a metric learning loss function. However, a single loss function often has certain limitations. Thus, we use deep multi-metric learning to address the problem and introduce three different losses for this problem, i.e., triplet loss, n-pair loss and angular loss. The three loss functions work in a cooperative way to train a feature extraction network equipped with Residual connections and squeeze-and-excitation attention. We conduct experiments on the large-scale \texttt{VoxCeleb2} dataset, which contains over a million utterances from over $6,000$ speakers, and the proposed deep neural network obtains an equal error rate of $3.48\%$, which is a very competitive result. Codes for both training and testing and pretrained models are available at \url{https://github.com/GreatJiweix/DmmlTiSV}, which is the first publicly available code repository for large-scale text-independent speaker verification with performance on par with the state-of-the-art systems.

eess.AS↗

arXiv2020

CCNet: Criss-Cross Attention for Semantic Segmentation

Contextual information is vital in visual understanding problems, such as semantic segmentation and object detection. We propose a Criss-Cross Network (CCNet) for obtaining full-image contextual information in a very effective and efficient way. Concretely, for each pixel, a novel criss-cross attention module harvests the contextual information of all the pixels on its criss-cross path. By taking a further recurrent operation, each pixel can finally capture the full-image dependencies. Besides, a category consistent loss is proposed to enforce the criss-cross attention module to produce more discriminative features. Overall, CCNet is with the following merits: 1) GPU memory friendly. Compared with the non-local block, the proposed recurrent criss-cross attention module requires 11x less GPU memory usage. 2) High computational efficiency. The recurrent criss-cross attention significantly reduces FLOPs by about 85% of the non-local block. 3) The state-of-the-art performance. We conduct extensive experiments on semantic segmentation benchmarks including Cityscapes, ADE20K, human parsing benchmark LIP, instance segmentation benchmark COCO, video segmentation benchmark CamVid. In particular, our CCNet achieves the mIoU scores of 81.9%, 45.76% and 55.47% on the Cityscapes test set, the ADE20K validation set and the LIP validation set respectively, which are the new state-of-the-art results. The source codes are available at \url{https://github.com/speedinghzl/CCNet}.

cs.CV↗

arXiv2020

IoU-aware Single-stage Object Detector for Accurate Localization

Due to the simpleness and high efficiency, single-stage object detectors have been widely applied in many computer vision applications . However, the low correlation between the classification score and localization accuracy of the predicted detections has severely hurt the localization accuracy of models. In this paper, IoU-aware single-stage object detector is proposed to solve this problem. Specifically, IoU-aware single-stage object detector predicts the IoU for each detected box. Then the classification score and predicted IoU are multiplied to compute the final detection confidence, which is more correlated with the localization accuracy. The detection confidence is then used as the input of the subsequent NMS and COCO AP computation, which will substantially improve the localization accuracy of models. Sufficient experiments on COCO and PASCAL VOC datasets demonstrate the effectiveness of IoU-aware single-stage object detector on improving model's localization accuracy. Without whistles and bells, the proposed method can substantially improve AP by $1.7\%\sim1.9\%$ and AP75 by $2.2\%\sim2.5\%$ on COCO \textit{test-dev}. On PASCAL VOC, the proposed method can substantially improve AP by $2.9\%\sim4.4\%$ and AP80, AP90 by $4.6\%\sim10.2\%$. Code is available here: {https://github.com/ShengkaiWu/IoU-aware-single-stage-object-detector}.

cs.CV↗

arXiv2020

Densely Connected Search Space for More Flexible Neural Architecture Search

Neural architecture search (NAS) has dramatically advanced the development of neural network design. We revisit the search space design in most previous NAS methods and find the number and widths of blocks are set manually. However, block counts and block widths determine the network scale (depth and width) and make a great influence on both the accuracy and the model cost (FLOPs/latency). In this paper, we propose to search block counts and block widths by designing a densely connected search space, i.e., DenseNAS. The new search space is represented as a dense super network, which is built upon our designed routing blocks. In the super network, routing blocks are densely connected and we search for the best path between them to derive the final architecture. We further propose a chained cost estimation algorithm to approximate the model cost during the search. Both the accuracy and model cost are optimized in DenseNAS. For experiments on the MobileNetV2-based search space, DenseNAS achieves 75.3% top-1 accuracy on ImageNet with only 361MB FLOPs and 17.9ms latency on a single TITAN-XP. The larger model searched by DenseNAS achieves 76.1% accuracy with only 479M FLOPs. DenseNAS further promotes the ImageNet classification accuracies of ResNet-18, -34 and -50-B by 1.5%, 0.5% and 0.3% with 200M, 600M and 680M FLOPs reduction respectively. The related code is available at https://github.com/JaminFong/DenseNAS.

cs.CV↗

arXiv2020

Fast Neural Network Adaptation via Parameter Remapping and Architecture Search

Deep neural networks achieve remarkable performance in many computer vision tasks. Most state-of-the-art (SOTA) semantic segmentation and object detection approaches reuse neural network architectures designed for image classification as the backbone, commonly pre-trained on ImageNet. However, performance gains can be achieved by designing network architectures specifically for detection and segmentation, as shown by recent neural architecture search (NAS) research for detection and segmentation. One major challenge though, is that ImageNet pre-training of the search space representation (a.k.a. super network) or the searched networks incurs huge computational cost. In this paper, we propose a Fast Neural Network Adaptation (FNA) method, which can adapt both the architecture and parameters of a seed network (e.g. a high performing manually designed backbone) to become a network with different depth, width, or kernels via a Parameter Remapping technique, making it possible to utilize NAS for detection/segmentation tasks a lot more efficiently. In our experiments, we conduct FNA on MobileNetV2 to obtain new networks for both segmentation and detection that clearly out-perform existing networks designed both manually and by NAS. The total computation cost of FNA is significantly less than SOTA segmentation/detection NAS approaches: 1737$\times$ less than DPC, 6.8$\times$ less than Auto-DeepLab and 7.4$\times$ less than DetNAS. The code is available at https://github.com/JaminFong/FNA.

cs.CV↗

arXiv2020

Deep High-Resolution Representation Learning for Visual Recognition

High-resolution representations are essential for position-sensitive vision problems, such as human pose estimation, semantic segmentation, and object detection. Existing state-of-the-art frameworks first encode the input image as a low-resolution representation through a subnetwork that is formed by connecting high-to-low resolution convolutions \emph{in series} (e.g., ResNet, VGGNet), and then recover the high-resolution representation from the encoded low-resolution representation. Instead, our proposed network, named as High-Resolution Network (HRNet), maintains high-resolution representations through the whole process. There are two key characteristics: (i) Connect the high-to-low resolution convolution streams \emph{in parallel}; (ii) Repeatedly exchange the information across resolutions. The benefit is that the resulting representation is semantically richer and spatially more precise. We show the superiority of the proposed HRNet in a wide range of applications, including human pose estimation, semantic segmentation, and object detection, suggesting that the HRNet is a stronger backbone for computer vision problems. All the codes are available at~{\url{https://github.com/HRNet}}.

cs.CV↗

arXiv2019

Diversity Transfer Network for Few-Shot Learning

Few-shot learning is a challenging task that aims at training a classifier for unseen classes with only a few training examples. The main difficulty of few-shot learning lies in the lack of intra-class diversity within insufficient training samples. To alleviate this problem, we propose a novel generative framework, Diversity Transfer Network (DTN), that learns to transfer latent diversities from known categories and composite them with support features to generate diverse samples for novel categories in feature space. The learning problem of the sample generation (i.e., diversity transfer) is solved via minimizing an effective meta-classification loss in a single-stage network, instead of the generative loss in previous works. Besides, an organized auxiliary task co-training over known categories is proposed to stabilize the meta-training process of DTN. We perform extensive experiments and ablation studies on three datasets, i.e., \emph{mini}ImageNet, CIFAR100 and CUB. The results show that DTN, with single-stage training and faster convergence speed, obtains the state-of-the-art results among the feature generation based few-shot learning methods. Code and supplementary material are available at: \texttt{https://github.com/Yuxin-CV/DTN}

cs.CV↗

arXiv2019

Object Detection in Video with Spatial-temporal Context Aggregation

Recent cutting-edge feature aggregation paradigms for video object detection rely on inferring feature correspondence. The feature correspondence estimation problem is fundamentally difficult due to poor image quality, motion blur, etc, and the results of feature correspondence estimation are unstable. To avoid the problem, we propose a simple but effective feature aggregation framework which operates on the object proposal-level. It learns to enhance each proposal's feature via modeling semantic and spatio-temporal relationships among object proposals from both within a frame and across adjacent frames. Experiments are carried out on the ImageNet VID dataset. Without any bells and whistles, our method obtains 80.3\% mAP on the ImageNet VID dataset, which is superior over the previous state-of-the-arts. The proposed feature aggregation mechanism improves the single frame Faster RCNN baseline by 5.8% mAP. Besides, under the setting of no temporal post-processing, our method outperforms the previous state-of-the-art by 1.4% mAP.

cs.CV↗