Feature coding has become increasingly important in scenarios where semantic representations rather than raw pixels are transmitted and stored. However, most existing methods are architecture-specific, targeting either CNNs or Transformers. This design limits their applicability in real-world scenarios where features from both architectures coexist. To address this gap, we introduce a new research problem: cross-architecture universal feature coding (CAUFC), which seeks to build a unified codec that can effectively compress features from heterogeneous architectures. To tackle this challenge, we propose a two-step distribution alignment method. First, we design the format alignment method that unifies CNN and Transformer features into a consistent 2D token format. Second, we propose the feature value alignment method that harmonizes statistical distributions via truncation and normalization. As a first attempt to study CAUFC, we evaluate our method on the image classification task. Experimental results demonstrate that our method achieves superior rate-accuracy trade-offs compared to the architecture-specific baseline. This work marks an initial step toward universal feature compression across heterogeneous model architectures.
Accurate prediction of molecular solubility is a cornerstone of early-stage drug discovery, yet conventional machine learning models face significant challenges due to limited labeled data and the high-dimensional nature of molecular descriptors. To address these issues, we propose LatMixSol, a novel latent space augmentation framework that combines autoencoder-based feature compression with guided interpolation to enrich training data. Our approach first encodes molecular descriptors into a low-dimensional latent space using a two-layer autoencoder. Spectral clustering is then applied to group chemically similar molecules, enabling targeted MixUp-style interpolation within clusters. Synthetic samples are generated by blending latent vectors of cluster members and decoding them back to the original feature space. Evaluated on the Huuskonen solubility benchmark, LatMixSol demonstrates consistent improvements across three of four gradient-boosted regressors (CatBoost, LightGBM, HistGradientBoosting), achieving RMSE reductions of 3.2-7.6% and R-squared increases of 0.5-1.5%. Notably, HistGradientBoosting shows the most significant enhancement with a 7.6% RMSE improvement. Our analysis confirms that cluster-guided latent space augmentation preserves chemical validity while expanding dataset diversity, offering a computationally efficient strategy to enhance predictive models in resource-constrained drug discovery pipelines.
Background: Federated Learning (FL) has emerged as a promising paradigm for training machine learning models while preserving data privacy. However, applying FL to Natural Language Processing (NLP) tasks presents unique challenges due to semantic heterogeneity across clients, vocabulary mismatches, and varying resource constraints on edge devices. Objectives: This paper introduces SEMFED, a novel semantic-aware resource-efficient federated learning framework specifically designed for heterogeneous NLP tasks. Methods: SEMFED incorporates three key innovations: (1) a semantic-aware client selection mechanism that balances semantic diversity with resource constraints, (2) adaptive NLP-specific model architectures tailored to device capabilities while preserving semantic information, and (3) a communication-efficient semantic feature compression technique that significantly reduces bandwidth requirements. Results: Experimental results on various NLP classification tasks demonstrate that SEMFED achieves an 80.5% reduction in communication costs while maintaining model accuracy above 98%, outperforming state-of-the-art FL approaches. Conclusion: SEMFED effectively manages heterogeneous client environments with varying computational resources, network reliability, and semantic data distributions, making it particularly suitable for real-world federated NLP deployments.
Curvilinear structure segmentation (CSS) is vital in various domains, including medical imaging, landscape analysis, industrial surface inspection, and plant analysis. While existing methods achieve high performance within specific domains, their generalizability is limited. On the other hand, large-scale models such as Segment Anything Model (SAM) exhibit strong generalization but are not optimized for curvilinear structures. Existing adaptations of SAM primarily focus on general object segmentation and lack specialized design for CSS tasks. To bridge this gap, we propose the Universal Curvilinear structure Segmentation (\textit{UCS}) model, which adapts SAM to CSS tasks while enhancing its generalization. \textit{UCS} features a novel encoder architecture integrating a pretrained SAM encoder with two innovations: a Sparse Adapter, strategically inserted to inherit the pre-trained SAM encoder's generalization capability while minimizing the number of fine-tuning parameters, and a Prompt Generation module, which leverages Fast Fourier Transform with a high-pass filter to generate curve-specific prompts. Furthermore, the \textit{UCS} incorporates a mask decoder that eliminates reliance on manual interaction through a dual-compression module: a Hierarchical Feature Compression module, which aggregates the outputs of the sampled encoder to enhance detail preservation, and a Guidance Feature Compression module, which extracts and compresses image-driven guidance features. Evaluated on a comprehensive multi-domain dataset, including an in-house dataset covering eight natural curvilinear structures, \textit{UCS} demonstrates state-of-the-art generalization and open-set segmentation performance across medical, engineering, natural, and plant imagery, establishing a new benchmark for universal CSS.
Nowadays, more and more video transmissions primarily aim at downstream machine vision tasks rather than humans. While widely deployed Human Visual System (HVS) oriented video coding standards like H.265/HEVC and H.264/AVC are efficient, they are not the optimal approaches for Video Coding for Machines (VCM) scenarios, leading to unnecessary bitrate expenditure. The academic and technical exploration within the VCM domain has led to the development of several strategies, and yet, conspicuous limitations remain in their adaptability for multi-task scenarios. To address the challenge, we propose a Transformable Video Feature Compression (TransVFC) framework. It offers a compress-then-transfer solution and includes a video feature codec and Feature Space Transform (FST) modules. In particular, the temporal redundancy of video features is squeezed by the codec through the scheme-based inter-prediction module. Then, the codec implements perception-guided conditional coding to minimize spatial redundancy and help the reconstructed features align with downstream machine perception.After that, the reconstructed features are transferred to new feature spaces for diverse downstream tasks by FST modules. To accommodate a new downstream task, it only requires training one lightweight FST module, avoiding retraining and redeploying the upstream codec and downstream task networks. Experiments show that TransVFC achieves high rate-task performance for diverse tasks of different granularities. We expect our work can provide valuable insights for video feature compression in multi-task scenarios. The codes are at https://github.com/Ws-Syx/TransVFC.
Efficient modal feature fusion strategy is the key to achieve accurate segmentation of brain glioma. However, due to the specificity of different MRI modes, it is difficult to carry out cross-modal fusion with large differences in modal features, resulting in the model ignoring rich feature information. On the other hand, the problem of multi-modal feature redundancy interaction occurs in parallel networks due to the proliferation of feature dimensions, further increase the difficulty of multi-modal feature fusion at the bottom end. In order to solve the above problems, we propose a noval complementary feature compression interaction network (CFCI-Net), which realizes the complementary fusion and compression interaction of multi-modal feature information with an efficient mode fusion strategy. Firstly, we propose a selective complementary feature fusion (SCFF) module, which adaptively fuses rich cross-modal feature information by complementary soft selection weights. Secondly, a modal feature compression interaction (MFCI) transformer is proposed to deal with the multi-mode fusion redundancy problem when the feature dimension surges. The MFCI transformer is composed of modal feature compression (MFC) and modal feature interaction (MFI) to realize redundancy feature compression and multi-mode feature interactive learning. %In MFI, we propose a hierarchical interactive attention mechanism based on multi-head attention. Evaluations on the BraTS2019 and BraTS2020 datasets demonstrate that CFCI-Net achieves superior results compared to state-of-the-art models. Code: https://github.com/CDmm0/CFCI-Net
With the rapid development of large multimodal models (LMMs), multimodal understanding applications are emerging. As most LMM inference requests originate from edge devices with limited computational capabilities, the predominant inference pipeline involves directly forwarding the input data to an edge server which handles all computations. However, this approach introduces high transmission latency due to limited uplink bandwidth of edge devices and significant computation latency caused by the prohibitive number of visual tokens, thus hindering delay-sensitive tasks and degrading user experience. To address this challenge, we propose a task-oriented feature compression (TOFC) method for multimodal understanding in a device-edge co-inference framework, where visual features are merged by clustering and encoded by a learnable and selective entropy model before feature projection. Specifically, we employ density peaks clustering based on K nearest neighbors to reduce the number of visual features, thereby minimizing both data transmission and computational complexity. Subsequently, a learnable entropy model with hyperprior is utilized to encode and decode merged features, further reducing transmission overhead. To enhance compression efficiency, multiple entropy models are adaptively selected based on the characteristics of the visual features, enabling a more accurate estimation of the probability distribution. Comprehensive experiments on seven visual question answering benchmarks validate the effectiveness of the proposed TOFC method. Results show that TOFC achieves up to 60% reduction in data transmission overhead and 50% reduction in system latency while maintaining identical task performance, compared with traditional image compression methods.
Fusing LiDAR point cloud features and image features in a homogeneous BEV space has been widely adopted for 3D object detection in autonomous driving. However, such methods are limited by the excessive compression of multi-modal features. While some works explore feature fusion in dense voxel spaces, they suffer from high computational costs and inefficiencies in query generation. To address these limitations, we propose a Dual-Domain Homogeneous Fusion network (DDHFusion), which leverages the complementary advantages of both BEV and voxel domains while mitigating their respective drawbacks. Specifically, we first transform image features into BEV and sparse voxel spaces using LSS and our proposed semantic-aware feature sampling module which can significantly reduces computational overhead by filtering unimportant voxels. For feature encoding, we design two networks for BEV and voxel feature fusion, incorporating novel cross-modal voxel and BEV Mamba blocks to resolve feature misalignment and enable efficient yet comprehensive scene perception. The output voxel features are injected into the BEV space to compensate for the loss of 3D details caused by height compression. For feature decoding, a progressive query generation module is implemented in the BEV domain to alleviate false negatives during query selection caused by feature compression and small object sizes. Finally, a progressive decoder can sequentially aggregate not only context-rich BEV features but also geometry-aware voxel features, ensuring more precise confidence prediction and bounding box regression. On the NuScenes dataset, DDHfusion achieves state-of-the-art performance, and further experiments demonstrate its superiority over other homogeneous fusion methods.
The emerging semantic compression has been receiving increasing research efforts most recently, capable of achieving high fidelity restoration during compression, even at extremely low bitrates. However, existing semantic compression methods typically combine standard pipelines with either pre-defined or high-dimensional semantics, thus suffering from deficiency in compression. To address this issue, we propose a novel hierarchical semantic compression (HSC) framework that purely operates within intrinsic semantic spaces from generative models, which is able to achieve efficient compression for consistent semantic restoration. More specifically, we first analyse the entropy models for the semantic compression, which motivates us to employ a hierarchical architecture based on a newly developed general inversion encoder. Then, we propose the feature compression network (FCN) and semantic compression network (SCN), such that the middle-level semantic feature and core semantics are hierarchically compressed to restore both accuracy and consistency of image semantics, via an entropy model progressively shared by channel-wise context. Experimental results demonstrate that the proposed HSC framework achieves the state-of-the-art performance on subjective quality and consistency for human vision, together with superior performances on machine vision tasks given compressed bitstreams. This essentially coincides with human visual system in understanding images, thus providing a new framework for future image/video compression paradigms. Our code shall be released upon acceptance.
2D image coding for machines (ICM) has achieved great success in coding efficiency, while less effort has been devoted to stereo image fields. To promote the efficiency of stereo image compression (SIC) and intelligent analysis, the stereo image coding for machines (SICM) is formulated and explored in this paper. More specifically, a machine vision-oriented stereo feature compression network (MVSFC-Net) is proposed for SICM, where the stereo visual features are effectively extracted, compressed, and transmitted for 3D visual task. To efficiently compress stereo visual features in MVSFC-Net, a stereo multi-scale feature compression (SMFC) module is designed to gradually transform sparse stereo multi-scale features into compact joint visual representations by removing spatial, inter-view, and cross-scale redundancies simultaneously. Experimental results show that the proposed MVSFC-Net obtains superior compression efficiency as well as 3D visual task performance, when compared with the existing ICM anchors recommended by MPEG and the state-of-the-art SIC method.
The pursuit of scaling up recommendation models confronts intrinsic tensions between expanding model capacity and preserving computational tractability. While prior studies have explored scaling laws for recommendation systems, their resource-intensive paradigms -- often requiring tens of thousands of A100 GPU hours -- remain impractical for most industrial applications. This work addresses a critical gap: achieving sustainable model scaling under strict computational budgets. We propose Climber, a resource-efficient recommendation framework comprising two synergistic components: the ASTRO model architecture for algorithmic innovation and the TURBO acceleration framework for engineering optimization. ASTRO (Adaptive Scalable Transformer for RecOmmendation) adopts two core innovations: (1) multi-scale sequence partitioning that reduces attention complexity from O(n^2d) to O(n^2d/Nb) via hierarchical blocks, enabling more efficient scaling with sequence length; (2) dynamic temperature modulation that adaptively adjusts attention scores for multimodal distributions arising from inherent multi-scenario and multi-behavior interactions. Complemented by TURBO (Two-stage Unified Ranking with Batched Output), a co-designed acceleration framework integrating gradient-aware feature compression and memory-efficient Key-Value caching, Climber achieves 5.15x throughput gains without performance degradation. Comprehensive offline experiments on multiple datasets validate that Climber exhibits a more ideal scaling curve. To our knowledge, this is the first publicly documented framework where controlled model scaling drives continuous online metric growth (12.19% overall lift) without prohibitive resource costs. Climber has been successfully deployed on Netease Cloud Music, one of China's largest music streaming platforms, serving tens of millions of users daily.
Recent advancements in multimodal large language models (MLLMs) have shown promising results, yet existing approaches struggle to effectively handle both temporal and spatial localization simultaneously. This challenge stems from two key issues: first, incorporating spatial-temporal localization introduces a vast number of coordinate combinations, complicating the alignment of linguistic and visual coordinate representations; second, encoding fine-grained temporal and spatial information during video feature compression is inherently difficult. To address these issues, we propose LLaVA-ST, a MLLM for fine-grained spatial-temporal multimodal understanding. In LLaVA-ST, we propose Language-Aligned Positional Embedding, which embeds the textual coordinate special token into the visual space, simplifying the alignment of fine-grained spatial-temporal correspondences. Additionally, we design the Spatial-Temporal Packer, which decouples the feature compression of temporal and spatial resolutions into two distinct point-to-region attention processing streams. Furthermore, we propose ST-Align dataset with 4.3M training samples for fine-grained spatial-temporal multimodal understanding. With ST-align, we present a progressive training pipeline that aligns the visual and textual feature through sequential coarse-to-fine stages.Additionally, we introduce an ST-Align benchmark to evaluate spatial-temporal interleaved fine-grained understanding tasks, which include Spatial-Temporal Video Grounding (STVG) , Event Localization and Captioning (ELC) and Spatial Video Grounding (SVG). LLaVA-ST achieves outstanding performance on 11 benchmarks requiring fine-grained temporal, spatial, or spatial-temporal interleaving multimodal understanding. Our code, data and benchmark will be released at Our code, data and benchmark will be released at https://github.com/appletea233/LLaVA-ST .
In recent years, vision language models (VLMs) have made significant advancements in video understanding. However, a crucial capability - fine-grained motion comprehension - remains under-explored in current benchmarks. To address this gap, we propose MotionBench, a comprehensive evaluation benchmark designed to assess the fine-grained motion comprehension of video understanding models. MotionBench evaluates models' motion-level perception through six primary categories of motion-oriented question types and includes data collected from diverse sources, ensuring a broad representation of real-world video content. Experimental results reveal that existing VLMs perform poorly in understanding fine-grained motions. To enhance VLM's ability to perceive fine-grained motion within a limited sequence length of LLM, we conduct extensive experiments reviewing VLM architectures optimized for video feature compression and propose a novel and efficient Through-Encoder (TE) Fusion method. Experiments show that higher frame rate inputs and TE Fusion yield improvements in motion understanding, yet there is still substantial room for enhancement. Our benchmark aims to guide and motivate the development of more capable video understanding models, emphasizing the importance of fine-grained motion comprehension. Project page: https://motion-bench.github.io .
Mobile edge computing (MEC) has empowered mobile devices (MDs) in supporting artificial intelligence (AI) applications through collaborative efforts with proximal MEC servers. Unfortunately, despite the great promise of device-edge cooperative AI inference, data privacy becomes an increasing concern. In this paper, we develop a privacy-aware multi-device cooperative edge inference system for classification tasks, which integrates a distributed bidding mechanism for the MEC server's computational resources. Intermediate feature compression is adopted as a principled approach to minimize data privacy leakage. To determine the bidding values and feature compression ratios in a distributed fashion, we formulate a decentralized partially observable Markov decision process (DEC-POMDP) model, for which, a multi-agent deep deterministic policy gradient (MADDPG)-based algorithm is developed. Simulation results demonstrate the effectiveness of the proposed algorithm in privacy-preserving cooperative edge inference. Specifically, given a sufficient level of data privacy protection, the proposed algorithm achieves 0.31-0.95% improvements in classification accuracy compared to the approach being agnostic to the wireless channel conditions. The performance is further enhanced by 1.54-1.67% by considering the difficulties of inference data.
Medical image colorization can greatly enhance the interpretability of the underlying imaging modality and provide insights into human anatomy. The objective of medical image colorization is to transfer a diverse spectrum of colors distributed across human anatomy from Cryosection data to source MRI data while retaining the structures of the MRI. To achieve this, we propose a novel architecture for structurally consistent color transfer to the source MRI data. Our architecture fuses segmentation semantics of Cryosection images for stable contextual colorization of various organs in MRI images. For colorization, we neither require precise registration between MRI and Cryosection images, nor segmentation of MRI images. Additionally, our architecture incorporates a feature compression-and-activation mechanism to capture organ-level global information and suppress noise, enabling the distinction of organ-specific data in MRI scans for more accurate and realistic organ-specific colorization. Our experiments demonstrate that our architecture surpasses the existing methods and yields better quantitative and qualitative results.
We present Meissonic, which elevates non-autoregressive masked image modeling (MIM) text-to-image to a level comparable with state-of-the-art diffusion models like SDXL. By incorporating a comprehensive suite of architectural innovations, advanced positional encoding strategies, and optimized sampling conditions, Meissonic substantially improves MIM's performance and efficiency. Additionally, we leverage high-quality training data, integrate micro-conditions informed by human preference scores, and employ feature compression layers to further enhance image fidelity and resolution. Our model not only matches but often exceeds the performance of existing models like SDXL in generating high-quality, high-resolution images. Extensive experiments validate Meissonic's capabilities, demonstrating its potential as a new standard in text-to-image synthesis. We release a model checkpoint capable of producing $1024 \times 1024$ resolution images.
Edge-cloud co-inference enables efficient deep neural network (DNN) deployment by splitting the architecture between an edge device and cloud server, crucial for resource-constraint edge devices. This approach requires balancing on-device computations and communication costs, often achieved through compressed intermediate feature transmission. Conventional DNN architectures require continuous data processing and floating point activations, leading to considerable energy consumption and increased feature sizes, thus raising transmission costs. This challenge motivates exploring binary, event-driven activations using spiking neural networks (SNNs), known for their extreme energy efficiency. In this research, we propose SpikeBottleNet, a novel architecture for edge-cloud co-inference systems that integrates a spiking neuron model to significantly reduce energy consumption on edge devices. A key innovation of our study is an intermediate feature compression technique tailored for SNNs for efficient feature transmission. This technique leverages a split computing approach to strategically place encoder-decoder bottleneck units within complex deep architectures like ResNet and MobileNet. Experimental results demonstrate that SpikeBottleNet achieves up to 256x bit compression in the final convolutional layer of ResNet, with minimal accuracy loss (0.16%). Additionally, our approach enhances edge device energy efficiency by up to 144x compared to the baseline BottleNet, making it ideal for resource-limited edge devices.
Traditional image codecs emphasize signal fidelity and human perception, often at the expense of machine vision tasks. Deep learning methods have demonstrated promising coding performance by utilizing rich semantic embeddings optimized for both human and machine vision. However, these compact embeddings struggle to capture fine details such as contours and textures, resulting in imperfect reconstructions. Furthermore, existing learning-based codecs lack scalability. To address these limitations, this paper introduces a content-adaptive diffusion model for scalable image compression. The proposed method encodes fine textures through a diffusion process, enhancing perceptual quality while preserving essential features for machine vision tasks. The approach employs a Markov palette diffusion model combined with widely used feature extractors and image generators, enabling efficient data compression. By leveraging collaborative texture-semantic feature extraction and pseudo-label generation, the method accurately captures texture information. A content-adaptive Markov palette diffusion model is then applied to represent both low-level textures and high-level semantic content in a scalable manner. This framework offers flexible control over compression ratios by selecting intermediate diffusion states, eliminating the need for retraining deep learning models at different operating points. Extensive experiments demonstrate the effectiveness of the proposed framework in both image reconstruction and downstream machine vision tasks such as object detection, segmentation, and facial landmark detection, achieving superior perceptual quality compared to state-of-the-art methods.
Deep trackers have proven success in visual tracking. Typically, these trackers employ optimally pre-trained deep networks to represent all diverse objects with multi-channel features from some fixed layers. The deep networks employed are usually trained to extract rich knowledge from massive data used in object classification and so they are capable to represent generic objects very well. However, these networks are too complex to represent a specific moving object, leading to poor generalization as well as high computational and memory costs. This paper presents a novel and general framework termed channel distillation to facilitate deep trackers. To validate the effectiveness of channel distillation, we take discriminative correlation filter (DCF) and ECO for example. We demonstrate that an integrated formulation can turn feature compression, response map generation, and model update into a unified energy minimization problem to adaptively select informative feature channels that improve the efficacy of tracking moving objects on the fly. Channel distillation can accurately extract good channels, alleviating the influence of noisy channels and generally reducing the number of channels, as well as adaptively generalizing to different channels and networks. The resulting deep tracker is accurate, fast, and has low memory requirements. Extensive experimental evaluations on popular benchmarks clearly demonstrate the effectiveness and generalizability of our framework.
Machine vision systems, which can efficiently manage extensive visual perception tasks, are becoming increasingly popular in industrial production and daily life. Due to the challenge of simultaneously obtaining accurate depth and texture information with a single sensor, multimodal data captured by cameras and LiDAR is commonly used to enhance performance. Additionally, cloud-edge cooperation has emerged as a novel computing approach to improve user experience and ensure data security in machine vision systems. This paper proposes a pioneering solution to address the feature compression problem in multimodal 3D object detection. Given a sparse tensor-based object detection network at the edge device, we introduce two modes to accommodate different application requirements: Transmission-Friendly Feature Compression (T-FFC) and Accuracy-Friendly Feature Compression (A-FFC). In T-FFC mode, only the output of the last layer of the network's backbone is transmitted from the edge device. The received feature is processed at the cloud device through a channel expansion module and two spatial upsampling modules to generate multi-scale features. In A-FFC mode, we expand upon the T-FFC mode by transmitting two additional types of features. These added features enable the cloud device to generate more accurate multi-scale features. Experimental results on the KITTI dataset using the VirConv-L detection network showed that T-FFC was able to compress the features by a factor of 6061 with less than a 3% reduction in detection performance. On the other hand, A-FFC compressed the features by a factor of about 901 with almost no degradation in detection performance. We also designed optional residual extraction and 3D object reconstruction modules to facilitate the reconstruction of detected objects. The reconstructed objects effectively reflected details of the original objects.