arXiv subjects

Jiajun Tang

Publications and source records attributed to Jiajun Tang.

17 recordsLinked to original sources

arXiv2026

VeraRetouch: A Lightweight Fully Differentiable Framework for Multi-Task Reasoning Photo Retouching

Reasoning photo retouching has gained significant traction, requiring models to analyze image defects, give reasoning processes, and execute precise retouching enhancements. However, existing approaches often rely on non-differentiable external software, creating optimization barriers and suffering from high parameter redundancy and limited generalization. To address these challenges, we propose VeraRetouch, a lightweight and fully differentiable framework for multi-task photo retouching. We employ a 0.5B Vision-Language Model (VLM) as the central intelligence to formulate retouching plans based on instructions and scene semantics. Furthermore, we develop a fully differentiable Retouch Renderer that replaces external tools, enabling direct end-to-end pixel-level training through decoupled control latents for lighting, global color, and specific color adjustments. To overcome data scarcity, we introduce AetherRetouch-1M+, the first million-scale dataset for professional retouching, constructed via a new inverse degradation workflow. Furthermore, we propose DAPO-AE, a reinforcement learning post-training strategy that enhances autonomous aesthetic cognition. Extensive experiments demonstrate that VeraRetouch achieves state-of-the-art performance across multiple benchmarks while maintaining a significantly smaller footprint, enabling mobile deployment. Our code and models are publicly available at https://github.com/OpenVeraTeam/VeraRetouch.

cs.CV↗

arXiv2026

Do AI Overviews Benefit Search Engines? An Ecosystem Perspective

The integration of AI Overviews into search engines enhances user experience but diverts traffic from content creators, potentially discouraging high-quality content creation and causing user attrition that undermines long-term search engine profit. To address this issue, we propose a game-theoretic model of creator competition with costly effort, characterize equilibrium behavior, and design two incentive mechanisms: a citation mechanism that references sources within an AI Overview, and a compensation mechanism that offers monetary rewards to creators. For both cases, we provide structural insights and near-optimal profit-maximizing mechanisms. Evaluations on real click data show that although AI Overviews harm long-term search engine profit, interventions based on our proposed mechanisms can increase long-term profit across a range of realistic scenarios, pointing toward a more sustainable trajectory for AI-enhanced search ecosystems.

cs.GT↗

arXiv2025

Towards Deeper Emotional Reflection: Crafting Affective Image Filters with Generative Priors

Social media platforms enable users to express emotions by posting text with accompanying images. In this paper, we propose the Affective Image Filter (AIF) task, which aims to reflect visually-abstract emotions from text into visually-concrete images, thereby creating emotionally compelling results. We first introduce the AIF dataset and the formulation of the AIF models. Then, we present AIF-B as an initial attempt based on a multi-modal transformer architecture. After that, we propose AIF-D as an extension of AIF-B towards deeper emotional reflection, effectively leveraging generative priors from pre-trained large-scale diffusion models. Quantitative and qualitative experiments demonstrate that AIF models achieve superior performance for both content consistency and emotional fidelity compared to state-of-the-art methods. Extensive user study experiments demonstrate that AIF models are significantly more effective at evoking specific emotions. Based on the presented results, we comprehensively discuss the value and potential of AIF models.

cs.CV↗

arXiv2025

No Redundancy, No Stall: Lightweight Streaming 3D Gaussian Splatting for Real-time Rendering

3D Gaussian Splatting (3DGS) enables high-quality rendering of 3D scenes and is getting increasing adoption in domains like autonomous driving and embodied intelligence. However, 3DGS still faces major efficiency challenges when faced with high frame rate requirements and resource-constrained edge deployment. To enable efficient 3DGS, in this paper, we propose LS-Gaussian, an algorithm/hardware co-design framework for lightweight streaming 3D rendering. LS-Gaussian is motivated by the core observation that 3DGS suffers from substantial computation redundancy and stalls. On one hand, in practical scenarios, high-frame-rate 3DGS is often applied in settings where a camera observes and renders the same scene continuously but from slightly different viewpoints. Therefore, instead of rendering each frame separately, LS-Gaussian proposes a viewpoint transformation algorithm that leverages inter-frame continuity for efficient sparse rendering. On the other hand, as different tiles within an image are rendered in parallel but have imbalanced workloads, frequent hardware stalls also slow down the rendering process. LS-Gaussian predicts the workload for each tile based on viewpoint transformation to enable more balanced parallel computation and co-designs a customized 3DGS accelerator to support the workload-aware mapping in real-time. Experimental results demonstrate that LS-Gaussian achieves 5.41x speedup over the edge GPU baseline on average and up to 17.3x speedup with the customized accelerator, while incurring only minimal visual quality degradation.

cs.AR↗

arXiv2025

Hybrid Mesh-Gaussian Representation for Efficient Indoor Scene Reconstruction

3D Gaussian splatting (3DGS) has demonstrated exceptional performance in image-based 3D reconstruction and real-time rendering. However, regions with complex textures require numerous Gaussians to capture significant color variations accurately, leading to inefficiencies in rendering speed. To address this challenge, we introduce a hybrid representation for indoor scenes that combines 3DGS with textured meshes. Our approach uses textured meshes to handle texture-rich flat areas, while retaining Gaussians to model intricate geometries. The proposed method begins by pruning and refining the extracted mesh to eliminate geometrically complex regions. We then employ a joint optimization for 3DGS and mesh, incorporating a warm-up strategy and transmittance-aware supervision to balance their contributions seamlessly.Extensive experiments demonstrate that the hybrid representation maintains comparable rendering quality and achieves superior frames per second FPS with fewer Gaussian primitives.

cs.CV↗

arXiv2025

PacTure: Efficient PBR Texture Generation on Packed Views with Visual Autoregressive Models

We present PacTure, a novel framework for generating physically-based rendering (PBR) material textures for an untextured 3D mesh from a text description. Existing 2D generation-based texturing approaches either generate textures sequentially from different views, resulting in long inference times and globally inconsistent textures, or adopt multi-view generation with cross-view attention to enhance global consistency, which, however, limits the resolution for each view. In response to these weaknesses, we first introduce view packing, a novel technique that significantly increases the effective resolution for each view during multi-view generation, without imposing additional inference cost. Unlike UV mapping, it preserves the spatial proximity essential for image generation and maintains full compatibility with current 2D generative models. To further reduce the inferencing cost, we enable fine-grained control and multi-domain generation within the next-scale prediction autoregressive framework, creating an efficient multi-view PBR generation backbone. Extensive experiments show that PacTure outperforms state-of-the-art methods in both quality and efficiency.

cs.CV↗

arXiv2025

Network Goodness-of-Fit for the block-model family

The block-model family has four popular network models (SBM, DCBM, MMSBM, and DCMM). A fundamental problem is, how well each of these models fits with real networks. We propose GoF-MSCORE as a new Goodness-of-Fit (GoF) metric for DCMM (the broadest one among the four), with two main ideas. The first is to use cycle count statistics as a general recipe for GoF. The second is a novel network fitting scheme. GoF-MSCORE is a flexible GoF approach, and we further extend it to SBM, DCBM, and MMSBM. This gives rise to a series of GoF metrics covering each of the four models in the block-model family. We show that for each of the four models, if the assumed model is correct, then the corresponding GoF metric converges to $N(0, 1)$ as the network sizes diverge. We also analyze the powers and show that these metrics are optimal in many settings. In comparison, many other GoF ideas face challenges: they may lack a parameter-free limiting null, or are non-optimal in power, or face an analytical hurdle. Note that a parameter-free limiting null is especially desirable as many network models have a large number of unknown parameters. The limiting nulls of our GoF metrics are always $N(0,1)$, which are parameter-free as desired. For 12 frequently-used real networks, we use the proposed GoF metrics to show that DCMM fits well with almost all of them. We also show that SBM, DCBM, and MMSBM do not fit well with many of these networks, especially when the networks are relatively large. To complement with our study on GoF, we also show that the DCMM is nearly as broad as the rank-$K$ network model. Based on these results, we recommend the DCMM as a promising model for undirected networks.

math.ST↗

arXiv2024

Simultaneous semiparametric inference for single-index models

In the common partially linear single-index model we establish a Bahadur representation for a smoothing spline estimator of all model parameters and use this result to prove the joint weak convergence of the estimator of the index link function at a given point, together with the estimators of the parametric regression coefficients. We obtain the surprising result that, despite of the nature of single-index models where the link function is evaluated at a linear combination of the index-coefficients, the estimator of the link function and the estimator of the index-coefficients are asymptotically independent. Our approach leverages a delicate analysis based on reproducing kernel Hilbert space and empirical process theory. We show that the smoothing spline estimator achieves the minimax optimal rate with respect to the $L^2$-risk and consider several statistical applications where joint inference on all model parameters is of interest. In particular, we develop a simultaneous confidence band for the link function and propose inference tools to investigate if the maximum absolute deviation between the (unknown) link function and a given function exceeds a given threshold. We also construct tests for joint hypotheses regarding model parameters which involve both the nonparametric and parametric components and propose novel multiplier bootstrap procedures to avoid the estimation of unknown asymptotic quantities.

math.ST↗

arXiv2024

New energy distances for statistical inference on infinite dimensional Hilbert spaces without moment conditions

For statistical inference on an infinite-dimensional Hilbert space $\H $ with no moment conditions we introduce a new class of energy distances on the space of probability measures on $\H$. The proposed distances consist of the integrated squared modulus of the corresponding difference of the characteristic functionals with respect to a reference probability measure on the Hilbert space. Necessary and sufficient conditions are established for the reference probability measure to be {\em characteristic}, the property that guarantees that the distance defines a metric on the space of probability measures on $\H$. We also use these results to define new distance covariances, which can be used to measure the dependence between the marginals of a two dimensional distribution of $\H^2$ without existing moments. On the basis of the new distances we develop statistical inference for Hilbert space valued data, which does not require any moment assumptions. As a consequence, our methods are robust with respect to heavy tails in finite dimensional data. In particular, we consider the problem of comparing the distributions of two samples and the problem of testing for independence and construct new minimax optimal tests for the corresponding hypotheses. We also develop aggregated (with respect to the reference measure) procedures for power enhancement and investigate the finite-sample properties by means of a simulation study.

math.ST↗

arXiv2024

Improved Algorithm and Bounds for Successive Projection

Given a $K$-vertex simplex in a $d$-dimensional space, suppose we measure $n$ points on the simplex with noise (hence, some of the observed points fall outside the simplex). Vertex hunting is the problem of estimating the $K$ vertices of the simplex. A popular vertex hunting algorithm is successive projection algorithm (SPA). However, SPA is observed to perform unsatisfactorily under strong noise or outliers. We propose pseudo-point SPA (pp-SPA). It uses a projection step and a denoise step to generate pseudo-points and feed them into SPA for vertex hunting. We derive error bounds for pp-SPA, leveraging on extreme value theory of (possibly) high-dimensional random vectors. The results suggest that pp-SPA has faster rates and better numerical performances than SPA. Our analysis includes an improved non-asymptotic bound for the original SPA, which is of independent interest.

cs.LG↗

arXiv2023

Efficient spectral broadening and few-cycle pulse generation with multiple thin water films

High-energy, few-cycle laser pulses are essential for numerous applications in the fields of ultrafast optics and strong-field physics, due to their ultrafast temporal resolution and high peak intensity. In this work, different from the traditional hollow-core fibers and multiple thin solid plates, we represent the first demonstration of the octave-spanning supercontinuum broadening by utilizing multiple ultrathin liquid films (MTLFs) as the nonlinear media. The continuum covers a range from 380 to 1050 nm, corresponding to a Fourier transform limit pulse width of 2.5 fs, when 35 fs Ti:sapphire laser pulse is applied on the MTLFs. The output pulses are compressed to 3.9 fs by employing chirped mirrors. Furthermore, a continuous high-order harmonic spectrum up to the 33rd order is realized by subjecting the compressed laser pulses to interact with Kr gas. The utilization of flowing water films eliminates permanent optical damage and enables wider and stronger spectrum broadening. Therefore, this MTLFs scheme provides new solutions for the generation of highly efficient femtosecond supercontinuum and nonlinear pulse compression, with potential applications in the fields of strong-field physics and attosecond science.

physics.optics↗

arXiv2022

An RKHS approach for pivotal inference in functional linear regression

We develop methodology for testing hypotheses regarding the slope function in functional linear regression for time series via a reproducing kernel Hilbert space approach. In contrast to most of the literature, which considers tests for the exact nullity of the slope function, we are interested in the null hypothesis that the slope function vanishes only approximately, where deviations are measured with respect to the $L^2$-norm. An asymptotically pivotal test is proposed, which does not require the estimation of nuisance parameters and long-run covariances. The key technical tools to prove the validity of our approach include a uniform Bahadur representation and a weak invariance principle for a sequential process of estimates of the slope function. Both scalar-on-function and function-on-function linear regression are considered and finite-sample methods for implementing our methodology are provided. We also illustrate the potential of our methods by means of a small simulation study and a data example.

math.ST↗

arXiv2021

Statistical inference for function-on-function linear regression

Function-on-function linear regression is important for understanding the relationship between the response and the predictor that are both functions. In this article, we propose a reproducing kernel Hilbert space approach to function-on-function linear regressionvia the penalised least square, regularized by the thin-plate spline smoothness penalty. The minimax optimal convergence rate of our estimator of the coefficient function is studied. We derive the Bahadur representation, which allows us to propose statistical inference methods using bootstrap and the convergence of Banach-valued random variables in the sup-norm. We illustrate our method and verify our theoretical results via simulated data experiments and a real data example.

math.ST↗

arXiv2021

DeRenderNet: Intrinsic Image Decomposition of Urban Scenes with Shape-(In)dependent Shading Rendering

We propose DeRenderNet, a deep neural network to decompose the albedo and latent lighting, and render shape-(in)dependent shadings, given a single image of an outdoor urban scene, trained in a self-supervised manner. To achieve this goal, we propose to use the albedo maps extracted from scenes in videogames as direct supervision and pre-compute the normal and shadow prior maps based on the depth maps provided as indirect supervision. Compared with state-of-the-art intrinsic image decomposition methods, DeRenderNet produces shadow-free albedo maps with clean details and an accurate prediction of shadows in the shape-independent shading, which is shown to be effective in re-rendering and improving the accuracy of high-level vision tasks for urban scenes.

cs.CV↗

arXiv2020

Complex Sequential Understanding through the Awareness of Spatial and Temporal Concepts

Understanding sequential information is a fundamental task for artificial intelligence. Current neural networks attempt to learn spatial and temporal information as a whole, limited their abilities to represent large scale spatial representations over long-range sequences. Here, we introduce a new modeling strategy called Semi-Coupled Structure (SCS), which consists of deep neural networks that decouple the complex spatial and temporal concepts learning. Semi-Coupled Structure can learn to implicitly separate input information into independent parts and process these parts respectively. Experiments demonstrate that a Semi-Coupled Structure can successfully annotate the outline of an object in images sequentially and perform video action recognition. For sequence-to-sequence problems, a Semi-Coupled Structure can predict future meteorological radar echo images based on observed images. Taken together, our results demonstrate that a Semi-Coupled Structure has the capacity to improve the performance of LSTM-like models on large scale sequential tasks.

cs.CV↗

arXiv2020

Asynchronous Interaction Aggregation for Action Detection

Understanding interaction is an essential part of video action detection. We propose the Asynchronous Interaction Aggregation network (AIA) that leverages different interactions to boost action detection. There are two key designs in it: one is the Interaction Aggregation structure (IA) adopting a uniform paradigm to model and integrate multiple types of interaction; the other is the Asynchronous Memory Update algorithm (AMU) that enables us to achieve better performance by modeling very long-term interaction dynamically without huge computation cost. We provide empirical evidence to show that our network can gain notable accuracy from the integrative interactions and is easy to train end-to-end. Our method reports the new state-of-the-art performance on AVA dataset, with 3.7 mAP gain (12.6% relative improvement) on validation split comparing to our strong baseline. The results on dataset UCF101-24 and EPIC-Kitchens further illustrate the effectiveness of our approach. Source code will be made public at: https://github.com/MVIG-SJTU/AlphAction .

cs.CV↗

arXiv2019

Three Branches: Detecting Actions With Richer Features

We present our three branch solutions for International Challenge on Activity Recognition at CVPR2019. This model seeks to fuse richer information of global video clip, short human attention and long-term human activity into a unified model. We have participated in two tasks: Task A, the Kinetics challenge and Task B, spatio-temporal action localization challenge. For Kinetics, we achieve 21.59% error rate. For the AVA challenge, our final model obtains 32.49% mAP on the test sets, which outperforms all submissions to the AVA challenge at CVPR 2018 for more than 10% mAP. As the future work, we will introduce human activity knowledge, which is a new dataset including key information of human activity.

cs.CV↗