arXiv subjects

Eduardo Pavez

Publications and source records attributed to Eduardo Pavez.

At least 19 recordsLinked to original sources

arXiv2026

Spend Bits Where Queries Look: KV Cache Vector Quantization with Attention-Preserving Transforms

Long-context LLM decoding reads the key-value (KV) cache at every step. Loading it takes longer than computing attention over it, so throughput is bandwidth-bound. Hence, reducing the cache size can raise both decoding speed and serving capacity. The challenge is to reduce cache size while preserving the attention products, keeping reconstruction cheap, and using a fixed per-token bit count. At two bits per element, the most competitive methods rely on orthogonal transforms. However, existing techniques are either data-oblivious or use the query statistics without deriving the transform from a distortion criterion. Moreover, they rely on transforms built on top of random or Hadamard rotations, which equalize variances across entries rather than compacting energy, and fixed-width scalar quantizers, which are suboptimal at low rates. In this paper, we formulate KV cache quantization as a transform coding problem in which distortion is the error in the attention products. We derive closed-form optimal transforms for keys and values from calibration statistics, under a high-resolution model. We show that the optimal key transform is not orthogonal and satisfies a generalized Parseval relation: the attention-aware distortion becomes mean-squared error (MSE) in the transform domain. Thus, we can use MSE-optimal vector quantizers applied directly to the transformed key coefficients. To meet the fixed-width layout requirement, we show that grouping coefficients into equal-volume partitions makes equal-size codebooks attain the variable-rate optimum under the same high-resolution model. At two bits per element, our method, termed NOVA-KV, recovers most of the long-context retrieval accuracy lost by scalar quantization methods at comparable throughput.

cs.LG↗

arXiv2026

Reduced-complexity Adaptive Loop Filtering via Input-dependent Graph Filters

Adaptive Loop Filtering is an important tool for suppressing compression artifacts in modern video codecs. In the enhanced compression model (ECM), a software test model used for experimenting with video coding tools beyond Versatile Video Coding, fixed filters are trained offline and achieve high signal adaptivity via a fine-grained gradient-based classifier, resulting in a large number of fixed filters that introduce redundancy and increased implementation complexity. Reducing this redundancy without compromising artifact suppression, therefore, remains a key challenge. This paper proposes an alternative graph-based fixed-filtering framework for adaptive loop filtering. By using a graph to encode pixel-intensity relationships, our approach captures local structural information more effectively than gradient-based classification alone. Fixed filters are learned as polynomial graph filters, enabling structurally similar local patterns to share common filtering behavior. Experimental results demonstrate that the proposed approach achieves a comparable performance to the ECM baseline while reducing the number of required filters by an order of magnitude.

eess.IV↗

arXiv2026

Motion Estimation Techniques for Volumetric Video Attribute Compression

Point cloud compression relies on techniques to compress both geometry and attributes. Motion-based approaches for dynamic solid point cloud geometry compression within the geometry-based point cloud compression (G-PCC) framework have achieved significant reductions in geometry rate. However, motion-based techniques for attribute compression remain underexplored, making it challenging to achieve significant reductions in the temporal redundancy of attributes. Firstly, this paper proposes a geometry-based inter-coding scheme to compress the attributes of dynamic solid point clouds. Secondly, a graph-based motion-estimation scheme for point-cloud attribute compression is proposed. Thirdly, an interpolation-free fractional-voxel motion estimation method is proposed to refine motion accuracy to fractional-voxel precision. Our experimental results on the MPEG point cloud dataset show that the proposed scheme outperforms G-PCC, GeS-TM, and V-PCC in lossless and lossy geometry conditions. We achieve average bitrate savings of $55.3\%$, $42.3\%$, and $16.5\%$ over G-PCC, GeS-TM, and V-PCC, respectively, under lossy-geometry conditions.

eess.IV↗

arXiv2026

GS-NFS: Bandwidth-adaptive Streaming of Dynamic Gaussian Splats and Point Clouds

Dynamic 3D Gaussian Splatting (3DGS) holds great promise as a 3D video streaming technology since it can represent complex 3D scenes with high fidelity. In this approach, every frame in a 3D video represents the environment as a collection of Gaussians with position and other attributes such as scale, rotation, opacity, and color. Frames capture fine details, permit views from any arbitrary perspective, but are an order of magnitude, or more, larger than 2D video frames. A line of recent work has explored how to compress dynamic 3DGS frames, but these approaches are often slow, in part because their compression techniques are not amenable to efficient acceleration. GS-NFS accelerates dynamic 3DGS compression and decompression on a GPU, to the point where it can encode and decode at full frame rate. It achieves this by developing novel GPU-based parallelizations of existing algorithms for encoding both positions and attributes of Gaussians. As a result, it is 1-2 orders of magnitude faster than the state-of-the-art in encoding and decoding a frame, while offering competitive compression performance and rendering quality.

cs.MM↗

arXiv2026

L2G-Net: Local to Global Spectral Graph Neural Networks via Cauchy Factorizations

Despite their theoretical advantages, spectral methods based on the graph Fourier transform (GFT) are seldom used in graph neural networks (GNNs) due to the cost of computing the eigenbasis and the lack of vertex-domain locality in the resulting representations. As a result, most GNNs rely on local approximations such as polynomial Laplacian filters or message passing, which limit their ability to model long-range dependencies. In this paper, we introduce an exact factorization of the GFT into operators acting on subgraphs, which are then combined via a sequence of Cauchy matrices. Building on this factorization, we propose a new class of spectral GNNs, termed L2G-Net (Local to Global Net). Unlike existing spectral methods, which are either fully global (when using the GFT) or local (when using polynomial filters), L2G-Net operates by processing the spectral representations of subgraphs and then combining them via structured matrices. Our algorithm avoids full eigendecompositions, exploiting graph topology to construct the factorization with quadratic complexity in the number of nodes, scaled by the maximum cut size between subgraphs. Experiments stressing long-range dependencies on large graphs show that L2G-Net scales to regimes out of reach for the standard GFT, and is competitive with state-of-the-art methods with orders of magnitude fewer learnable parameters.

cs.LG↗

arXiv2026

FaSST: Fast Sparsifying Secondary Transform

Data-dependent secondary transforms, which aim to decorrelate coefficients of a separable primary transform, can improve residual coding efficiency; however, their deployment is often constrained by computational complexity. Recent video codecs use variants of the low-frequency non-separable transform (LFNST), which discards some high-frequency secondary transform coefficients, limiting achievable coding gains. Moreover, existing data-dependent secondary transforms lack explicit rate-distortion (RD) optimal design criteria. In this work, we propose a framework for designing low-complexity data-dependent secondary transforms, termed Fast Sparsifying Secondary Transforms (FaSSTs). Our approach approximates data-driven sparse orthonormal transforms (SOTs) by factorizing them into a sequence of Givens rotations. The rotations are efficiently determined using an alternating minimization strategy combined with an approximate Givens factorization procedure. Our method adapts the number of rotations based on the prediction mode, further reducing computational complexity. We design mode-dependent secondary transforms for intra-prediction residuals in AV2 using FaSST. Experimental results show that mode-adaptive FaSST matches the RD performance of LFNST while reducing the number of computations by 83.67%. Moreover, by avoiding fixed-coefficient truncation, FaSST achieves up to 1.80% BD-rate savings relative to LFNST while operating at 66.24% lower complexity.

eess.IV↗

arXiv2026

Rate-Distortion Optimization for Ensembles of Non-Reference Metrics

Non-reference metrics (NRMs) can assess the visual quality of images and videos without a reference, making them well-suited for the evaluation of user-generated content. Nonetheless, rate-distortion optimization (RDO) in video coding is still mainly driven by full-reference metrics, such as the sum of squared errors, which treat the input as an ideal target. A way to incorporate NRMs into RDO is through linearization (LNRM), where the gradient of the NRM with respect to the input guides bit allocation. While this strategy improves the quality predicted by some metrics, we show that it can yield limited gains or degradations when evaluated with other NRMs. We argue that NRMs are highly non-linear predictors with locally unstable gradients that can compromise the quality of the linearization; furthermore, optimizing a single metric may exploit model-specific biases that do not generalize across quality estimators. Motivated by this observation, we extend the LNRM framework to optimize ensembles of NRMs and, to further improve robustness, we introduce a smoothing-based formulation that stabilizes NRM gradients prior to linearization. Our framework is well-suited to hybrid codecs, and we advocate for its use with overfitted codecs, where it avoids iterative evaluations and backpropagation of neural network-based NRMs, reducing encoder complexity relative to direct NRM optimization. We validate the proposed approach on AVC and Cool-chic, using the YouTube UGC dataset. Experiments demonstrate consistent bitrate savings across multiple NRMs with no decoder complexity overhead and, for Cool-chic, a substantial reduction in encoding runtime compared to direct NRM optimization.

eess.IV↗

arXiv2026

Wrapper-Aware Rate-Distortion Optimization in Feature Coding for Machines

Feature coding for machines (FCM) is a lossy compression paradigm for split-inference. The transmitter encodes the outputs of the first part of a neural network before sending them to the receiver for completing the inference. Practical FCM methods ``sandwich'' a traditional codec between pre- and post-processing neural networks, called wrappers, to make features easier to compress using video codecs. Since traditional codecs are non-differentiable, the wrappers are trained using a proxy codec, which is later replaced by a standard codec after training. These codecs perform rate-distortion optimization (RDO) based on the sum of squared errors (SSE). Because the RDO does not consider the post-processing wrapper, the inner codec can invest bits in preserving information that the post-processing later discards. In this paper, we modify the bit-allocation in the inner codec via a wrapper-aware weighted SSE metric. To make wrapper-aware RDO (WA-RDO) practical for FCM, we propose: 1) temporal reuse of weights across a group of pictures and 2) fixed, architecture- and task-dependent weights trained offline. Under MPEG test conditions, our methods implemented on HEVC match the VVC-based FCM state-of-the-art, effectively bridging a codec generation gap with minimal runtime overhead relative to SSE-RDO HEVC.

eess.IV↗

arXiv2025

INT-DTT+: Low-Complexity Data-Dependent Transforms for Video Coding

Discrete trigonometric transforms (DTTs), such as the DCT-2 and the DST-7, are widely used in video codecs for their balance between coding performance and computational efficiency. In contrast, data-dependent transforms, such as the Karhunen-Loève transform (KLT) and graph-based separable transforms (GBSTs), offer better energy compaction but lack symmetries that can be exploited to reduce computational complexity. This paper bridges this gap by introducing a general framework to design low-complexity data-dependent transforms. Our approach builds on DTT+, a family of GBSTs derived from rank-one updates of the DTT graphs, which can adapt to signal statistics while retaining a structure amenable to fast computation. We first propose a graph learning algorithm for DTT+ that estimates the rank-one updates for rows and column graphs jointly, capturing the statistical properties of the overall block. Then, we exploit the progressive structure of DTT+ to decompose the kernel into a base DTT and a structured Cauchy matrix. By leveraging low-complexity integer DTTs and sparsifying the Cauchy matrix, we construct an integer approximation to DTT+, termed INT-DTT+. This approximation significantly reduces both computational and memory complexities with respect to the separable KLT with minimal performance loss. We validate our approach in the context of mode-dependent transforms for the VVC standard, following a rate-distortion optimized transform (RDOT) design approach. Integrated into the explicit multiple transform selection (MTS) framework of VVC in a rate-distortion optimization setup, INT-DTT+ achieves more than 3% BD-rate savings over the VVC MTS baseline, with complexity comparable to the integer DCT-2 once the base DTT coefficients are available.

eess.IV↗

arXiv2025

Avoiding Quality Saturation in UGC Compression Using Denoised References

Video-sharing platforms must re-encode large volumes of noisy user-generated content (UGC) to meet streaming demands. However, conventional codecs, which aim to minimize the mean squared error (MSE) between the compressed and input videos, can cause quality saturation (QS) when applied to UGC, i.e., increasing the bitrate preserves input artifacts without improving visual quality. A direct approach to solve this problem is to detect QS by repeatedly evaluating a non-reference metric (NRM) on videos compressed with multiple codec parameters, which is inefficient. In this paper, we re-frame UGC compression and QS detection from the lens of noisy source coding theory: rather than using a NRM, we compute the MSE with respect to the denoised UGC, which serves as an alternative reference (D-MSE). Unlike MSE measured between the UGC input and the compressed UGC, D-MSE saturates at non-zero values as bitrates increase, a phenomenon we term distortion saturation (DS). Since D-MSE can be computed at the block level in the transform domain, we can efficiently detect D-MSE without coding and decoding with various parameters. We propose two methods for DS detection: distortion saturation detection (DSD), which relies on an input-dependent threshold derived from the D-MSE of the input UGC, and rate-distortion saturation detection (RDSD), which estimates the Lagrangian at the saturation point using a low-complexity compression method. Both methods work as a pre-processing step that can help standard-compliant codecs avoid QS in UGC compression. Experiments with AVC show that preventing encoding in the saturation region, i.e., avoiding encoding at QPs that result in QS according to our methods, achieves BD-rate savings of 8%-20% across multiple different NRMs, compared to a naïve baseline that encodes at the given input QP while ignoring QS.

eess.IV↗

arXiv2025

Region-Adaptive Learned Hierarchical Encoding for 3D Gaussian Splatting Data

We introduce Region-Adaptive Learned Hierarchical Encoding (RALHE) for 3D Gaussian Splatting (3DGS) data. While 3DGS has recently become popular for novel view synthesis, the size of trained models limits its deployment in bandwidth-constrained applications such as volumetric media streaming. To address this, we propose a learned hierarchical latent representation that builds upon the principles of "overfitted" learned image compression (e.g., Cool-Chic and C3) to efficiently encode 3DGS attributes. Unlike images, 3DGS data have irregular spatial distributions of Gaussians (geometry) and consist of multiple attributes (signals) defined on the irregular geometry. Our codec is designed to account for these differences between images and 3DGS. Specifically, we leverage the octree structure of the voxelized 3DGS geometry to obtain a hierarchical multi-resolution representation. Our approach overfits latents to each Gaussian attribute under a global rate constraint. These latents are decoded independently through a lightweight decoder network. To estimate the bitrate during training, we employ an autoregressive probability model that leverages octree-derived contexts from the 3D point structure. The multi-resolution latents, decoder, and autoregressive entropy coding networks are jointly optimized for each Gaussian attribute. Experiments demonstrate that the proposed RALHE compression framework achieves a rendering PSNR gain of up to 2dB at low bitrates (less than 1 MB) compared to the baseline 3DGS compression methods.

eess.IV↗

arXiv2025

Image Coding for Machines via Feature-Preserving Rate-Distortion Optimization

Many images and videos are primarily processed by computer vision algorithms, involving only occasional human inspection. When this content requires compression before processing, e.g., in distributed applications, coding methods must optimize for both visual quality and downstream task performance. We first show theoretically that an approach to reduce the effect of compression for a given task loss is to perform rate-distortion optimization (RDO) using the distance between features, obtained from the original and the decoded images, as a distortion metric. However, optimizing directly such a rate-distortion objective is computationally impractical because it requires iteratively encoding and decoding the entire image-plus feature evaluation-for each possible coding configuration. We address this problem by simplifying the RDO formulation to make the distortion term computable using block-based encoders. We first apply Taylor's expansion to the feature extractor, recasting the feature distance as a quadratic metric involving the Jacobian matrix of the neural network. Then, we replace the linearized metric with a block-wise approximation, which we call input-dependent squared error (IDSE). To make the metric computable, we approximate IDSE using sketches of the Jacobian. The resulting loss can be evaluated block-wise in the transform domain and combined with the sum of squared errors (SSE) to address both visual quality and computer vision performance. Simulations with AVC and HEVC across multiple feature extractors and downstream networks show up to 17 % bit-rate savings for the same task accuracy compared to RDO based on SSE, with no decoder complexity overhead and a small (7.86 %) encoder complexity increase.

eess.IV↗

arXiv2025

Adaptive Voxelization for Transform coding of 3D Gaussian splatting data

We present a novel compression framework for 3D Gaussian splatting (3DGS) data that leverages transform coding tools originally developed for point clouds. Contrary to existing 3DGS compression methods, our approach can produce compressed 3DGS models at multiple bitrates in a computationally efficient way. Point cloud voxelization is a discretization technique that point cloud codecs use to improve coding efficiency while enabling the use of fast transform coding algorithms. We propose an adaptive voxelization algorithm tailored to 3DGS data, to avoid the inefficiencies introduced by uniform voxelization used in point cloud codecs. We ensure the positions of larger volume Gaussians are represented at high resolution, as these significantly impact rendering quality. Meanwhile, a low-resolution representation is used for dense regions with smaller Gaussians, which have a relatively lower impact on rendering quality. This adaptive voxelization approach significantly reduces the number of Gaussians and the bitrate required to encode the 3DGS data. After voxelization, many Gaussians are moved or eliminated. Thus, we propose to fine-tune/recolor the remaining 3DGS attributes with an initialization that can reduce the amount of retraining required. Experimental results on pre-trained datasets show that our proposed compression framework outperforms existing methods.

eess.IV↗

arXiv2025

Rate-Distortion Optimization with Non-Reference Metrics for UGC Compression

Service providers must encode a large volume of noisy videos to meet the demand for user-generated content (UGC) in online video-sharing platforms. However, low-quality UGC challenges conventional codecs based on rate-distortion optimization (RDO) with full-reference metrics (FRMs). While effective for pristine videos, FRMs drive codecs to preserve artifacts when the input is degraded, resulting in suboptimal compression. A more suitable approach used to assess UGC quality is based on non-reference metrics (NRMs). However, RDO with NRMs as a measure of distortion requires an iterative workflow of encoding, decoding, and metric evaluation, which is computationally impractical. This paper overcomes this limitation by linearizing the NRM around the uncompressed video. The resulting cost function enables block-wise bit allocation in the transform domain by estimating the alignment of the quantization error with the gradient of the NRM. To avoid large deviations from the input, we add sum of squared errors (SSE) regularization. We derive expressions for both the SSE regularization parameter and the Lagrangian, akin to the relationship used for SSE-RDO. Experiments with images and videos show bitrate savings of more than 30\% over SSE-RDO using the target NRM, with no decoder complexity overhead and minimal encoder complexity increase.

eess.IV↗

arXiv2025

Joint Optimization of Primary and Secondary Transforms Using Rate-Distortion Optimized Transform Design

Data-dependent transforms are increasingly being incorporated into next-generation video coding systems such as AVM, a codec under development by the Alliance for Open Media (AOM), and VVC. To circumvent the computational complexities associated with implementing non-separable data-dependent transforms, combinations of separable primary transforms and non-separable secondary transforms have been studied and integrated into video coding standards. These codecs often utilize rate-distortion optimized transforms (RDOT) to ensure that the new transforms complement existing transforms like the DCT and the ADST. In this work, we propose an optimization framework for jointly designing primary and secondary transforms from data through a rate-distortion optimized clustering. Primary transforms are assumed to follow a path-graph model, while secondary transforms are non-separable. We empirically evaluate our proposed approach using AVM residual data and demonstrate that 1) the joint clustering method achieves lower total RD cost in the RDOT design framework, and 2) jointly optimized separable path-graph transforms (SPGT) provide better coding efficiency compared to separable KLTs obtained from the same data.

eess.IV↗

arXiv2024

Towards joint graph learning and sampling set selection from data

We explore the problem of sampling graph signals in scenarios where the graph structure is not predefined and must be inferred from data. In this scenario, existing approaches rely on a two-step process, where a graph is learned first, followed by sampling. More generally, graph learning and graph signal sampling have been studied as two independent problems in the literature. This work provides a foundational step towards jointly optimizing the graph structure and sampling set. Our main contribution, Vertex Importance Sampling (VIS), is to show that the sampling set can be effectively determined from the vertex importance (node weights) obtained from graph learning. We further propose Vertex Importance Sampling with Repulsion (VISR), a greedy algorithm where spatially -separated "important" nodes are selected to ensure better reconstruction. Empirical results on simulated data show that sampling using VIS and VISR leads to competitive reconstruction performance and lower complexity than the conventional two-step approach of graph learning followed by graph sampling.

cs.LG↗

arXiv2024

Color-Guided Flying Pixel Correction in Depth Images

We present a novel method to correct flying pixels within data captured by Time-of-flight (ToF) sensors. Flying pixel (FP) artifacts occur when signals from foreground and background objects reach the same sensor pixel, leading to a confident yet incorrect depth estimation in space - floating between two objects. Commercial RGB-D cameras have a complementary setup consisting of ToF sensors to capture depth in addition to RGB cameras. We propose a novel method to correct FPs by leveraging the aligned RGB and depth image in such RGB-D cameras to estimate the true depth values of FPs. Our method defines a 3D neighborhood around each point, representing a "field of view" that mirrors the acquisition process of ToF cameras. We propose a two-step iterative correction algorithm in which the FPs are first identified. Then, we estimate the true depth value of FPs by solving a least-squares optimization problem. Experimental results show that our proposed algorithm estimates the depth value of FPs as accurately as other algorithms in the literature.

eess.IV↗

arXiv2024

Graph-based Scalable Sampling of 3D Point Cloud Attributes

3D Point clouds (PCs) are commonly used to represent 3D scenes. They can have millions of points, making subsequent downstream tasks such as compression and streaming computationally expensive. PC sampling (selecting a subset of points) can be used to reduce complexity. Existing PC sampling algorithms focus on preserving geometry features and often do not scale to handle large PCs. In this work, we develop scalable graph-based sampling algorithms for PC color attributes, assuming the full geometry is available. Our sampling algorithms are optimized for a signal reconstruction method that minimizes the graph Laplacian quadratic form. We first develop a global sampling algorithm that can be applied to PCs with millions of points by exploiting sparsity and sampling rate adaptive parameter selection. Further, we propose a block-based sampling strategy where each block is sampled independently. We show that sampling the corresponding sub-graphs with optimally chosen self-loop weights (node weights) will produce a sampling set that approximates the results of global sampling while reducing complexity by an order of magnitude. Our empirical results on two large PC datasets show that our algorithms outperform the existing fast PC subsampling techniques (uniform and geometry feature preserving random sampling) by 2dB. Our algorithm is up to 50 times faster than existing graph signal sampling algorithms while providing better reconstruction accuracy. Finally, we illustrate the efficacy of PC attribute sampling within a compression scenario, showing that pre-compression sampling of PC attributes can lower the bitrate by 11% while having minimal effect on reconstruction.

eess.IV↗