arXiv subjects

Xiangyu Chen

Publications and source records attributed to Xiangyu Chen.

121 records · Page 7Linked to original sources

arXiv2021

Audio-Visual Transformer Based Crowd Counting

Crowd estimation is a very challenging problem. The most recent study tries to exploit auditory information to aid the visual models, however, the performance is limited due to the lack of an effective approach for feature extraction and integration. The paper proposes a new audiovisual multi-task network to address the critical challenges in crowd counting by effectively utilizing both visual and audio inputs for better modalities association and productive feature extraction. The proposed network introduces the notion of auxiliary and explicit image patch-importance ranking (PIR) and patch-wise crowd estimate (PCE) information to produce a third (run-time) modality. These modalities (audio, visual, run-time) undergo a transformer-inspired cross-modality co-attention mechanism to finally output the crowd estimate. To acquire rich visual features, we propose a multi-branch structure with transformer-style fusion in-between. Extensive experimental evaluations show that the proposed scheme outperforms the state-of-the-art networks under all evaluation settings with up to 33.8% improvement. We also analyze and compare the vision-only variant of our network and empirically demonstrate its superiority over previous approaches.

cs.CV↗

arXiv2021

Few-Shot Learning by Integrating Spatial and Frequency Representation

Human beings can recognize new objects with only a few labeled examples, however, few-shot learning remains a challenging problem for machine learning systems. Most previous algorithms in few-shot learning only utilize spatial information of the images. In this paper, we propose to integrate the frequency information into the learning model to boost the discrimination ability of the system. We employ Discrete Cosine Transformation (DCT) to generate the frequency representation, then, integrate the features from both the spatial domain and frequency domain for classification. The proposed strategy and its effectiveness are validated with different backbones, datasets, and algorithms. Extensive experiments demonstrate that the frequency information is complementary to the spatial representations in few-shot classification. The classification accuracy is boosted significantly by integrating features from both the spatial and frequency domains in different few-shot learning tasks.

cs.CV↗

arXiv2021

Design and Implementation of MIMO Transmission Based on Dual-Polarized Reconfigurable Intelligent Surface

Multiple-input multiple-output (MIMO) signaling is one of the key technologies of current mobile communication systems. However, the complex and expensive radio frequency (RF) chains have always limited the increase of MIMO scale. In this paper, we propose a MIMO transmission architecture based on a dual-polarized reconfigurable intelligent surface (RIS), which can directly achieve modulation and transmission of multichannel signals without the need for conventional RF chains. Compared with previous works, the proposed architecture can improve the integration of RIS-based transmission systems. A prototype of the dual-polarized RIS-based MIMO transmission system is built and the experimental results confirm the feasibility of the proposed architecture. The dual-polarized RIS-based MIMO transmission architecture provides a promising solution for realizing low-cost ultra-massive MIMO towards future networks.

cs.IT↗

arXiv2021

HDRUNet: Single Image HDR Reconstruction with Denoising and Dequantization

Most consumer-grade digital cameras can only capture a limited range of luminance in real-world scenes due to sensor constraints. Besides, noise and quantization errors are often introduced in the imaging process. In order to obtain high dynamic range (HDR) images with excellent visual quality, the most common solution is to combine multiple images with different exposures. However, it is not always feasible to obtain multiple images of the same scene and most HDR reconstruction methods ignore the noise and quantization loss. In this work, we propose a novel learning-based approach using a spatially dynamic encoder-decoder network, HDRUNet, to learn an end-to-end mapping for single image HDR reconstruction with denoising and dequantization. The network consists of a UNet-style base network to make full use of the hierarchical multi-scale information, a condition network to perform pattern-specific modulation and a weighting network for selectively retaining information. Moreover, we propose a Tanh_L1 loss function to balance the impact of over-exposed values and well-exposed values on the network learning. Our method achieves the state-of-the-art performance in quantitative comparisons and visual quality. The proposed HDRUNet model won the second place in the single frame track of NITRE2021 High Dynamic Range Challenge.

eess.IV↗

arXiv2021

Improving the List Decoding Version of the Cyclically Equivariant Neural Decoder

The cyclically equivariant neural decoder was recently proposed in [Chen-Ye, International Conference on Machine Learning, 2021] to decode cyclic codes. In the same paper, a list decoding procedure was also introduced for two widely used classes of cyclic codes -- BCH codes and punctured Reed-Muller (RM) codes. While the list decoding procedure significantly improves the Frame Error Rate (FER) of the cyclically equivariant neural decoder, the Bit Error Rate (BER) of the list decoding procedure is even worse than the unique decoding algorithm when the list size is small. In this paper, we propose an improved version of the list decoding algorithm for BCH codes and punctured RM codes. Our new proposal significantly reduces the BER while maintaining the same (in some cases even smaller) FER. More specifically, our new decoder provides up to $2$dB gain over the previous list decoder when measured by BER, and the running time of our new decoder is $15\%$ smaller. Code available at https://github.com/improvedlistdecoder/code

cs.IT↗

arXiv2021

Low-Precision Reinforcement Learning: Running Soft Actor-Critic in Half Precision

Low-precision training has become a popular approach to reduce compute requirements, memory footprint, and energy consumption in supervised learning. In contrast, this promising approach has not yet enjoyed similarly widespread adoption within the reinforcement learning (RL) community, partly because RL agents can be notoriously hard to train even in full precision. In this paper we consider continuous control with the state-of-the-art SAC agent and demonstrate that a naïve adaptation of low-precision methods from supervised learning fails. We propose a set of six modifications, all straightforward to implement, that leaves the underlying agent and its hyperparameters unchanged but improves the numerical stability dramatically. The resulting modified SAC agent has lower memory and compute requirements while matching full-precision rewards, demonstrating that low-precision training can substantially accelerate state-of-the-art RL without parameter tuning.

cs.LG↗

arXiv2021

Cyclically Equivariant Neural Decoders for Cyclic Codes

Neural decoders were introduced as a generalization of the classic Belief Propagation (BP) decoding algorithms, where the Trellis graph in the BP algorithm is viewed as a neural network, and the weights in the Trellis graph are optimized by training the neural network. In this work, we propose a novel neural decoder for cyclic codes by exploiting their cyclically invariant property. More precisely, we impose a shift invariant structure on the weights of our neural decoder so that any cyclic shift of inputs results in the same cyclic shift of outputs. Extensive simulations with BCH codes and punctured Reed-Muller (RM) codes show that our new decoder consistently outperforms previous neural decoders when decoding cyclic codes. Finally, we propose a list decoding procedure that can significantly reduce the decoding error probability for BCH codes and punctured RM codes. For certain high-rate codes, the gap between our list decoder and the Maximum Likelihood decoder is less than $0.1$dB. Code available at https://github.com/cyclicallyneuraldecoder/CyclicallyEquivariantNeuralDecoders

cs.IT↗

arXiv2020

Robotic grasp detection using a novel two-stage approach

Recently, deep learning has been successfully applied to robotic grasp detection. Based on convolutional neural networks (CNNs), there have been lots of end-to-end detection approaches. But end-to-end approaches have strict requirements for the dataset used for training the neural network models and it's hard to achieve in practical use. Therefore, we proposed a two-stage approach using particle swarm optimizer (PSO) candidate estimator and CNN to detect the most likely grasp. Our approach achieved an accuracy of 92.8% on the Cornell Grasp Dataset, which leaped into the front ranks of the existing approaches and is able to run at real-time speeds. After a small change of the approach, we can predict multiple grasps per object in the meantime so that an object can be grasped in a variety of ways.

cs.RO↗

arXiv2020

Wireless Communications with Reconfigurable Intelligent Surface: Path Loss Modeling and Experimental Measurement

Reconfigurable intelligent surfaces (RISs) comprised of tunable unit cells have recently drawn significant attention due to their superior capability in manipulating electromagnetic waves. In particular, RIS-assisted wireless communications have the great potential to achieve significant performance improvement and coverage enhancement in a cost-effective and energy-efficient manner, by properly programming the reflection coefficients of the unit cells of RISs. In this paper, free-space path loss models for RIS-assisted wireless communications are developed for different scenarios by studying the physics and electromagnetic nature of RISs. The proposed models, which are first validated through extensive simulation results, reveal the relationships between the free-space path loss of RIS-assisted wireless communications and the distances from the transmitter/receiver to the RIS, the size of the RIS, the near-field/far-field effects of the RIS, and the radiation patterns of antennas and unit cells. In addition, three fabricated RISs (metasurfaces) are utilized to further corroborate the theoretical findings through experimental measurements conducted in a microwave anechoic chamber. The measurement results match well with the modeling results, thus validating the proposed free-space path loss models for RIS, which may pave the way for further theoretical studies and practical applications in this field.

eess.SP↗

arXiv2020

Train in Germany, Test in The USA: Making 3D Object Detectors Generalize

In the domain of autonomous driving, deep learning has substantially improved the 3D object detection accuracy for LiDAR and stereo camera data alike. While deep networks are great at generalization, they are also notorious to over-fit to all kinds of spurious artifacts, such as brightness, car sizes and models, that may appear consistently throughout the data. In fact, most datasets for autonomous driving are collected within a narrow subset of cities within one country, typically under similar weather conditions. In this paper we consider the task of adapting 3D object detectors from one dataset to another. We observe that naively, this appears to be a very challenging task, resulting in drastic drops in accuracy levels. We provide extensive experiments to investigate the true adaptation challenges and arrive at a surprising conclusion: the primary adaptation hurdle to overcome are differences in car sizes across geographic areas. A simple correction based on the average car size yields a strong correction of the adaptation gap. Our proposed method is simple and easily incorporated into most 3D object detection frameworks. It provides a first baseline for 3D object detection adaptation across countries, and gives hope that the underlying problem may be more within grasp than one may have hoped to believe. Our code is available at https://github.com/cxy1997/3D_adapt_auto_driving.

cs.CV↗

arXiv2020

Transferable Active Grasping and Real Embodied Dataset

Grasping in cluttered scenes is challenging for robot vision systems, as detection accuracy can be hindered by partial occlusion of objects. We adopt a reinforcement learning (RL) framework and 3D vision architectures to search for feasible viewpoints for grasping by the use of hand-mounted RGB-D cameras. To overcome the disadvantages of photo-realistic environment simulation, we propose a large-scale dataset called Real Embodied Dataset (RED), which includes full-viewpoint real samples on the upper hemisphere with amodal annotation and enables a simulator that has real visual feedback. Based on this dataset, a practical 3-stage transferable active grasping pipeline is developed, that is adaptive to unseen clutter scenes. In our pipeline, we propose a novel mask-guided reward to overcome the sparse reward issue in grasping and ensure category-irrelevant behavior. The grasping pipeline and its possible variants are evaluated with extensive experiments both in simulation and on a real-world UR-5 robotic arm.

cs.RO↗

arXiv2020

Combating noisy labels by agreement: A joint training method with co-regularization

Deep Learning with noisy labels is a practically challenging problem in weakly supervised learning. The state-of-the-art approaches "Decoupling" and "Co-teaching+" claim that the "disagreement" strategy is crucial for alleviating the problem of learning with noisy labels. In this paper, we start from a different perspective and propose a robust learning paradigm called JoCoR, which aims to reduce the diversity of two networks during training. Specifically, we first use two networks to make predictions on the same mini-batch data and calculate a joint loss with Co-Regularization for each training example. Then we select small-loss examples to update the parameters of both two networks simultaneously. Trained by the joint loss, these two networks would be more and more similar due to the effect of Co-Regularization. Extensive experimental results on corrupted data from benchmark datasets including MNIST, CIFAR-10, CIFAR-100 and Clothing1M demonstrate that JoCoR is superior to many state-of-the-art approaches for learning with noisy labels.

cs.CV↗

arXiv2019

Weakly Aligned Cross-Modal Learning for Multispectral Pedestrian Detection

Multispectral pedestrian detection has shown great advantages under poor illumination conditions, since the thermal modality provides complementary information for the color image. However, real multispectral data suffers from the position shift problem, i.e. the color-thermal image pairs are not strictly aligned, making one object has different positions in different modalities. In deep learning based methods, this problem makes it difficult to fuse the feature maps from both modalities and puzzles the CNN training. In this paper, we propose a novel Aligned Region CNN (AR-CNN) to handle the weakly aligned multispectral data in an end-to-end way. Firstly, we design a Region Feature Alignment (RFA) module to capture the position shift and adaptively align the region features of the two modalities. Secondly, we present a new multimodal fusion method, which performs feature re-weighting to select more reliable features and suppress the useless ones. Besides, we propose a novel RoI jitter strategy to improve the robustness to unexpected shift patterns of different devices and system settings. Finally, since our method depends on a new kind of labelling: bounding boxes that match each modality, we manually relabel the KAIST dataset by locating bounding boxes in both modalities and building their relationships, providing a new KAIST-Paired Annotation. Extensive experimental validations on existing datasets are performed, demonstrating the effectiveness and robustness of the proposed method. Code and data are available at https://github.com/luzhang16/AR-CNN.

cs.CV↗