arXiv subjects

Ruoyu Yang

Publications and source records attributed to Ruoyu Yang.

7 recordsLinked to original sources

arXiv2026

Decoder-Guided Lossy Contour Coding Via Anchor Refinement

Object contours serve as compact structural priors for many receiver-side vision tasks such as image super-resolution, edge-conditioned generation, and machine vision. When such tasks are deployed over a bandwidth-limited channel, the sender transmits the high-quality object contour as structural side information to guide reconstruction at the receiver, while-to save bandwidth-only a low-quality reference such as a downsampled image or base-layer reconstruction is delivered. As a result, the decoder can already extract a coarse contour from this reference at no transmission cost, creating an encoder-decoder asymmetry: the fine contour must be coded and sent, yet a free coarse version is available at the decoder. This asymmetry is ignored by existing contour codecs such as JBIG2 and chain coding, which are lossless, symmetric, and offer no rate-distortion control, leading to high bitrates. In this paper, we propose a coarse-to-fine contour coding framework that models a high-quality contour as a structured geometric refinement of the decoder-available coarse contour. The encoder extracts ordered anchors along the fine contour and performs adaptive anchor skipping under a distortion constraint. The decoder then reconstructs the contour by using the coarse prior to guide anchor connectivity. This formulation enables lossy contour compression with an explicit rate-distortion trade-off. Experiments show 54.5%-66.9% bitrate reduction over methods without decoder-side guidance, and up to 5 times savings over JBIG2, while preserving high geometric accuracy.

eess.IV↗

arXiv2026

Uni-LVC: A Unified Method for Intra- and Inter-Mode Learned Video Compression

Recent advances in learned video compression (LVC) have led to significant performance gains, with codecs such as DCVC-RT surpassing the H.266/VVC low-delay mode in compression efficiency. However, existing LVCs still exhibit key limitations: they often require separate models for intra and inter coding modes, and their performance degrades when temporal references are unreliable. To address this, we introduce Uni-LVC, a unified LVC method that supports both intra and inter coding with low-delay and random-access in a single model. Building on a strong intra-codec, Uni-LVC formulates inter-coding as intra-coding conditioned on temporal information extracted from reference frames. We design an efficient cross-attention adaptation module that integrates temporal cues, enabling seamless support for both unidirectional (low-delay) and bidirectional (random-access) prediction modes. A reliability-aware classifier is proposed to selectively scale the temporal cues, making Uni-LVC behave closer to intra coding when references are unreliable. We further propose a multistage training strategy to facilitate adaptive learning across various coding modes. Extensive experiments demonstrate that Uni-LVC achieves superior rate-distortion performance in intra and inter configurations while maintaining comparable computational efficiency.

eess.IV↗

arXiv2025

Frequency-Domain Characterization of Load Demand from Electrified Highways

Electrified roadways (ER) equipped with dynamic wireless power transfer (DWPT) capabilities can patently extend the driving range and reduce the battery size of electric vehicles (EVs). However, due to the spatial arrangement of the transmitter coils in the ER, the DWPT load exhibits frequency content that could excite power system frequency dynamics. In this context, this work aims to study the spectrum of DWPT loads under different traffic conditions. Under simplifying assumptions, we develop statistical models to identify the location and relative magnitude of DWPT load harmonics. Our analysis reveals that the fundamental frequency depends on ER coil spacing and average EV speed. In the worst-case yet unlikely scenario that EVs move in a synchronized fashion, the amplitude of harmonics scales with the EV count. On the contrary, when EVs move freely, harmonics scale with the square root of the EV count. Platoon formations can accentuate harmonics. The spectral content around harmonics decreases in magnitude and increases in bandwidth with the harmonic index. The load of a single EV moving at a time-varying speed can be modeled as a frequency-modulated (FM) signal. Despite the simplifying assumptions, the derived models offer valuable insights for ER planners and grid operators. Dynamic simulations of a modified WECC model with DWPT loads synthesized from realistic EV trajectories and ER specifications corroborate some of these insights.

eess.SY↗

arXiv2022

Reverse Back Propagation to Make Full Use of Derivative

The development of the back-propagation algorithm represents a landmark in neural networks. We provide an approach that conducts the back-propagation again to reverse the traditional back-propagation process to optimize the input loss at the input end of a neural network for better effects without extra costs during the inference time. Then we further analyzed its principles and advantages and disadvantages, reformulated the weight initialization strategy for our method. And experiments on MNIST, CIFAR10, and CIFAR100 convinced our approaches could adapt to a larger range of learning rate and learn better than vanilla back-propagation.

cs.LG↗

arXiv2022

Sharp error estimate of variable time-step IMEX BDF2 scheme for parabolic integro-differential equations with initial singularity arising in finance

The recently developed technique of DOC kernels has been a great success in the stability and convergence analysis for BDF2 scheme with variable time steps. However, such an analysis technique seems not directly applicable to problems with initial singularity. In the numerical simulations of solutions with initial singularity, variable time-steps schemes like the graded mesh are always adopted to achieve the optimal convergence, whose first adjacent time-step ratio may become pretty large so that the acquired restriction is not satisfied. In this paper, we revisit the variable time-step implicit-explicit two-step backward differentiation formula (IMEX BDF2) scheme presented in [W. Wang, Y. Chen and H. Fang, \emph{SIAM J. Numer. Anal.}, 57 (2019), pp. 1289-1317] to compute the partial integro-differential equations (PIDEs) with initial singularity. We obtain the sharp error estimate under a mild restriction condition of adjacent time-step ratios $r_{k}: =τ_{k}/τ_{k-1} \; (k\geq 3) < r_{\max} = 4.8645 $ and a much mild requirement on the first ratio, i.e., $r_2>0$. This leads to the validation of our analysis of the variable time-step IMEX BDF2 scheme when the initial singularity is dealt by a simple strategy, i.e., the graded mesh $t_k=T(k/N)^γ$. In this situation, the convergence of order $\mathcal{O}(N^{-\min\{2,γα\}})$ is achieved with $N$ and $α$ respectively representing the total mesh points and indicating the regularity of the exact solution. This is, the optical convergence will be achieved by taking $γ_{\text{opt}}=2/α$. Numerical examples are provided to demonstrate our theoretical analysis.

math.NA↗

arXiv2021

Feature Boosting, Suppression, and Diversification for Fine-Grained Visual Classification

Learning feature representation from discriminative local regions plays a key role in fine-grained visual classification. Employing attention mechanisms to extract part features has become a trend. However, there are two major limitations in these methods: First, they often focus on the most salient part while neglecting other inconspicuous but distinguishable parts. Second, they treat different part features in isolation while neglecting their relationships. To handle these limitations, we propose to locate multiple different distinguishable parts and explore their relationships in an explicit way. In this pursuit, we introduce two lightweight modules that can be easily plugged into existing convolutional neural networks. On one hand, we introduce a feature boosting and suppression module that boosts the most salient part of feature maps to obtain a part-specific representation and suppresses it to force the following network to mine other potential parts. On the other hand, we introduce a feature diversification module that learns semantically complementary information from the correlated part-specific representations. Our method does not need bounding boxes/part annotations and can be trained end-to-end. Extensive experimental results show that our method achieves state-of-the-art performances on several benchmark fine-grained datasets. Source code is available at https://github.com/chaomaer/FBSD.

cs.CV↗

arXiv2021

Learning Granularity-Aware Convolutional Neural Network for Fine-Grained Visual Classification

Locating discriminative parts plays a key role in fine-grained visual classification due to the high similarities between different objects. Recent works based on convolutional neural networks utilize the feature maps taken from the last convolutional layer to mine discriminative regions. However, the last convolutional layer tends to focus on the whole object due to the large receptive field, which leads to a reduced ability to spot the differences. To address this issue, we propose a novel Granularity-Aware Convolutional Neural Network (GA-CNN) that progressively explores discriminative features. Specifically, GA-CNN utilizes the differences of the receptive fields at different layers to learn multi-granularity features, and it exploits larger granularity information based on the smaller granularity information found at the previous stages. To further boost the performance, we introduce an object-attentive module that can effectively localize the object given a raw image. GA-CNN does not need bounding boxes/part annotations and can be trained end-to-end. Extensive experimental results show that our approach achieves state-of-the-art performances on three benchmark datasets.

cs.CV↗