Bin Li

Jiahao Li, Bin Li, Yan Lu

For neural video codec, it is critical, yet challenging, to design an efficient entropy model which can accurately predict the probability distribution of the quantized latent representation. However, most existing video codecs directly use the ready-made entropy model from image codec to encode the residual or motion, and do not fully leverage the spatial-temporal characteristics in video. To this end, this paper proposes a powerful entropy model which efficiently captures both spatial and temporal dependencies. In particular, we introduce the latent prior which exploits the correlation among the latent representation to squeeze the temporal redundancy. Meanwhile, the dual spatial prior is proposed to reduce the spatial redundancy in a parallel-friendly manner. In addition, our entropy model is also versatile. Besides estimating the probability distribution, our entropy model also generates the quantization step at spatial-channel-wise. This content-adaptive quantization mechanism not only helps our codec achieve the smooth rate adjustment in single model but also improves the final rate-distortion performance by dynamic bit allocation. Experimental results show that, powered by the proposed entropy model, our neural codec can achieve 18.2% bitrate saving on UVG dataset when compared with H.266 (VTM) using the highest compression ratio configuration. It makes a new milestone in the development of neural video codec. The codes are at https://github.com/microsoft/DCVC.

Guo-Hua Wang, Jiahao Li, Bin Li et al.

Neural image compression has surpassed state-of-the-art traditional codecs (H.266/VVC) for rate-distortion (RD) performance, but suffers from large complexity and separate models for different rate-distortion trade-offs. In this paper, we propose an Efficient single-model Variable-bit-rate Codec (EVC), which is able to run at 30 FPS with 768x512 input images and still outperforms VVC for the RD performance. By further reducing both encoder and decoder complexities, our small model even achieves 30 FPS with 1920x1080 input images. To bridge the performance gap between our different capacities models, we meticulously design the mask decay, which transforms the large model's parameters into the small model automatically. And a novel sparsity regularization loss is proposed to mitigate shortcomings of $L_p$ regularization. Our algorithm significantly narrows the performance gap by 50% and 30% for our medium and small models, respectively. At last, we advocate the scalable encoder for neural image compression. The encoding complexity is dynamic to meet different latency requirements. We propose decaying the large encoder multiple times to reduce the residual representation progressively. Both mask decay and residual representation learning greatly improve the RD performance of our scalable encoder. Our code is at https://github.com/microsoft/DCVC.

Jiahao Li, Bin Li, Yan Lu

For any video codecs, the coding efficiency highly relies on whether the current signal to be encoded can find the relevant contexts from the previous reconstructed signals. Traditional codec has verified more contexts bring substantial coding gain, but in a time-consuming manner. However, for the emerging neural video codec (NVC), its contexts are still limited, leading to low compression ratio. To boost NVC, this paper proposes increasing the context diversity in both temporal and spatial dimensions. First, we guide the model to learn hierarchical quality patterns across frames, which enriches long-term and yet high-quality temporal contexts. Furthermore, to tap the potential of optical flow-based coding framework, we introduce a group-based offset diversity where the cross-group interaction is proposed for better context mining. In addition, this paper also adopts a quadtree-based partition to increase spatial context diversity when encoding the latent representation in parallel. Experiments show that our codec obtains 23.5% bitrate saving over previous SOTA NVC. Better yet, our codec has surpassed the under-developing next generation traditional codec/ECM in both RGB and YUV420 colorspaces, in terms of PSNR. The codes are at https://github.com/microsoft/DCVC.

Zhiyu Jin, Xuli Shen, Bin Li et al.

Diffusion models (DMs) have recently gained attention with state-of-the-art performance in text-to-image synthesis. Abiding by the tradition in deep learning, DMs are trained and evaluated on the images with fixed sizes. However, users are demanding for various images with specific sizes and various aspect ratio. This paper focuses on adapting text-to-image diffusion models to handle such variety while maintaining visual fidelity. First we observe that, during the synthesis, lower resolution images suffer from incomplete object portrayal, while higher resolution images exhibit repetitively disordered presentation. Next, we establish a statistical relationship indicating that attention entropy changes with token quantity, suggesting that models aggregate spatial information in proportion to image resolution. The subsequent interpretation on our observations is that objects are incompletely depicted due to limited spatial information for low resolutions, while repetitively disorganized presentation arises from redundant spatial information for high resolutions. From this perspective, we propose a scaling factor to alleviate the change of attention entropy and mitigate the defective pattern observed. Extensive experimental results validate the efficacy of the proposed scaling factor, enabling models to achieve better visual effects, image quality, and text alignment. Notably, these improvements are achieved without additional training or fine-tuning techniques.

Mingzhao Yang, Shangchao Su, Bin Li et al.

In recent years, the attention towards One-Shot Federated Learning (OSFL) has been driven by its capacity to minimize communication. With the development of the diffusion model (DM), several methods employ the DM for OSFL, utilizing model parameters, image features, or textual prompts as mediums to transfer the local client knowledge to the server. However, these mediums often require public datasets or the uniform feature extractor, significantly limiting their practicality. In this paper, we propose FedDEO, a Description-Enhanced One-Shot Federated Learning Method with DMs, offering a novel exploration of utilizing the DM in OSFL. The core idea of our method involves training local descriptions on the clients, serving as the medium to transfer the knowledge of the distributed clients to the server. Firstly, we train local descriptions on the client data to capture the characteristics of client distributions, which are then uploaded to the server. On the server, the descriptions are used as conditions to guide the DM in generating synthetic datasets that comply with the distributions of various clients, enabling the training of the aggregated model. Theoretical analyses and sufficient quantitation and visualization experiments on three large-scale real-world datasets demonstrate that through the training of local descriptions, the server is capable of generating synthetic datasets with high quality and diversity. Consequently, with advantages in communication and privacy protection, the aggregated model outperforms compared FL or diffusion-based OSFL methods and, on some clients, outperforms the performance ceiling of centralized training.

Yueqi Zeng, Ziqiang Li, Pengfei Xia et al.

With the boom in the natural language processing (NLP) field these years, backdoor attacks pose immense threats against deep neural network models. However, previous works hardly consider the effect of the poisoning rate. In this paper, our main objective is to reduce the number of poisoned samples while still achieving a satisfactory Attack Success Rate (ASR) in text backdoor attacks. To accomplish this, we propose an efficient trigger word insertion strategy in terms of trigger word optimization and poisoned sample selection. Extensive experiments on different datasets and models demonstrate that our proposed method can significantly improve attack effectiveness in text classification tasks. Remarkably, our approach achieves an ASR of over 90% with only 10 poisoned samples in the dirty-label setting and requires merely 1.5% of the training data in the clean-label setting.

Dongdong Lin, Yue Li, Benedetta Tondi et al.

The rapid proliferation of Deep Neural Networks (DNNs) is driving a surge in model watermarking technologies, as the trained models themselves constitute valuable intellectual property. Existing watermarking approaches primarily focus on modifying model parameters or altering sampling behaviors. However, with the emergence of increasingly large models, improving the efficiency of watermark embedding becomes essential to manage increasing computational demands. Prioritizing efficiency not only optimizes resource utilization, making the watermarking process more applicable for large models, but also mitigates potential degradation of model performance. In this paper, we propose an efficient watermarking method for Latent Diffusion Models (LDMs) based on Low-Rank Adaptation (LoRA). The core idea is to introduce trainable low-rank parameters into the frozen LDM to embed watermark, thereby preserving the integrity of the original model weights. Furthermore, a dynamic loss weight scheduler is designed to adaptively balance the objectives of generative quality and watermark fidelity, enabling the model to achieve effective watermark embedding with minimal impact on quality of the generated images. Experimental results show that the proposed method ensures fast and accurate watermark embedding and a high quality of the generated images, at the same time maintaining a level of robustness aligned - in some cases superior - with state-of-the-art approaches. Moreover, the method generalizes well across different datasets and base LDMs. Codes are available at: https://github.com/MrDongdongLin/EW-LoRA.

Zhiqing Zhang, Tianyong Liu, Guojia Fan et al.

Accurate segmentation of 3D clinical medical images is critical in the diagnosis and treatment of spinal diseases. However, the inherent complexity of spinal anatomy and uncertainty inherent in current imaging technologies, poses significant challenges for semantic segmentation of spinal images. Although convolutional neural networks (CNNs) and Transformer-based models have made some progress in spinal segmentation, their limitations in handling long-range dependencies hinder further improvements in segmentation accuracy.To address these challenges, we introduce a residual visual Mamba layer to effectively capture and model the deep semantic features and long-range spatial dependencies of 3D spinal data. To further enhance the structural semantic understanding of the vertebrae, we also propose a novel spinal shape prior module that captures specific anatomical information of the spine from medical images, significantly enhancing the model's ability to extract structural semantic information of the vertebrae. Comparative and ablation experiments on two datasets demonstrate that SpineMamba outperforms existing state-of-the-art models. On the CT dataset, the average Dice similarity coefficient for segmentation reaches as high as 94.40, while on the MR dataset, it reaches 86.95. Notably, compared to the renowned nnU-Net, SpineMamba achieves superior segmentation performance, exceeding it by up to 2 percentage points. This underscores its accuracy, robustness, and excellent generalization capabilities.

Wufei Ma, Jiahao Li, Bin Li et al.

Deep learning-based video compression is a challenging task, and many previous state-of-the-art learning-based video codecs use optical flows to exploit the temporal correlation between successive frames and then compress the residual error. Although these two-stage models are end-to-end optimized, the epistemic uncertainty in the motion estimation and the aleatoric uncertainty from the quantization operation lead to errors in the intermediate representations and introduce artifacts in the reconstructed frames. This inherent flaw limits the potential for higher bit rate savings. To address this issue, we propose an uncertainty-aware video compression model that can effectively capture the predictive uncertainty with deep ensembles. Additionally, we introduce an ensemble-aware loss to encourage the diversity among ensemble members and investigate the benefits of incorporating adversarial training in the video compression task. Experimental results on 1080p sequences show that our model can effectively save bits by more than 20% compared to DVC Pro.

Linfeng Qi, Zhaoyang Jia, Jiahao Li et al.

Most existing approaches for image and video compression perform transform coding in the pixel space to reduce redundancy. However, due to the misalignment between the pixel-space distortion and human perception, such schemes often face the difficulties in achieving both high-realism and high-fidelity at ultra-low bitrate. To solve this problem, we propose \textbf{G}enerative \textbf{L}atent \textbf{C}oding (\textbf{GLC}) models for image and video compression, termed GLC-image and GLC-Video. The transform coding of GLC is conducted in the latent space of a generative vector-quantized variational auto-encoder (VQ-VAE). Compared to the pixel-space, such a latent space offers greater sparsity, richer semantics and better alignment with human perception, and show its advantages in achieving high-realism and high-fidelity compression. To further enhance performance, we improve the hyper prior by introducing a spatial categorical hyper module in GLC-image and a spatio-temporal categorical hyper module in GLC-video. Additionally, the code-prediction-based loss function is proposed to enhance the semantic consistency. Experiments demonstrate that our scheme shows high visual quality at ultra-low bitrate for both image and video compression. For image compression, GLC-image achieves an impressive bitrate of less than $0.04$ bpp, achieving the same FID as previous SOTA model MS-ILLM while using $45\%$ fewer bitrate on the CLIC 2020 test set. For video compression, GLC-video achieves 65.3\% bitrate saving over PLVC in terms of DISTS.

Zhibin Wen, Bin Li

The goal of Multilingual Visual Answer Localization (MVAL) is to locate a video segment that answers a given multilingual question. Existing methods either focus solely on visual modality or integrate visual and subtitle modalities. However, these methods neglect the audio modality in videos, consequently leading to incomplete input information and poor performance in the MVAL task. In this paper, we propose a unified Audio-Visual-Textual Span Localization (AVTSL) method that incorporates audio modality to augment both visual and textual representations for the MVAL task. Specifically, we integrate features from three modalities and develop three predictors, each tailored to the unique contributions of the fused modalities: an audio-visual predictor, a visual predictor, and a textual predictor. Each predictor generates predictions based on its respective modality. To maintain consistency across the predicted results, we introduce an Audio-Visual-Textual Consistency module. This module utilizes a Dynamic Triangular Loss (DTL) function, allowing each modality's predictor to dynamically learn from the others. This collaborative learning ensures that the model generates consistent and comprehensive answers. Extensive experiments show that our proposed method outperforms several state-of-the-art (SOTA) methods, which demonstrates the effectiveness of the audio modality.

Jinming Liu, Zhaoyang Jia, Jiahao Li et al.

The increasing deployment of powerful Multimodal Large Language Models (MLLMs), typically hosted on cloud platforms, urgently requires effective compression techniques to efficiently transmit signal inputs (e.g., images, videos) from edge devices with minimal bandwidth usage. However, conventional image codecs are optimized for fidelity to serve the Human Visual System (HVS) and ill-suited for MLLMs, in which diverse downstream tasks are jointly considered. In this paper, we first systematically analyze the impact of compression artifacts on several mainstream MLLMs. We find that: Compression distortion unevenly impacts different-level image features, leading to varying effects on MLLMs' downstream tasks depending on their feature-level reliance. Motivated by this discovery, we propose an image Codec TAilored to MLLMs (CoTAM) designed to adaptively protect multi-level features and suit different demands of downstream tasks. The encoder leverages CLIP's shallow-layer attention to generate an importance map for bit allocation, preserving critical semantic regions. Concurrently, the decoder integrates a lightweight adapter with a multi-level loss function to ensure the faithful reconstruction both of low-level details and high-level semantic context for robust synthesis of cross-level features. Extensive experiments validate that our method achieves up to 35.99\% bitrate saving while maintaining the same performance on the MLLM tasks, outperforming previous SOTA neural codecs.

Jiahao Li, Bin Li, Yan Lu

Most of the existing neural video compression methods adopt the predictive coding framework, which first generates the predicted frame and then encodes its residue with the current frame. However, as for compression ratio, predictive coding is only a sub-optimal solution as it uses simple subtraction operation to remove the redundancy across frames. In this paper, we propose a deep contextual video compression framework to enable a paradigm shift from predictive coding to conditional coding. In particular, we try to answer the following questions: how to define, use, and learn condition under a deep video compression framework. To tap the potential of conditional coding, we propose using feature domain context as condition. This enables us to leverage the high dimension context to carry rich information to both the encoder and the decoder, which helps reconstruct the high-frequency contents for higher video quality. Our framework is also extensible, in which the condition can be flexibly designed. Experiments show that our method can significantly outperform the previous state-of-the-art (SOTA) deep video compression methods. When compared with x265 using veryslow preset, we can achieve 26.0% bitrate saving for 1080P standard test videos.

Benedetta Tondi, Andrea Costranzo, Dequ Huang et al.

Estimating the primary quantization matrix of double JPEG compressed images is a problem of relevant importance in image forensics since it allows to infer important information about the past history of an image. In addition, the inconsistencies of the primary quantization matrices across different image regions can be used to localize splicing in double JPEG tampered images. Traditional model-based approaches work under specific assumptions on the relationship between the first and second compression qualities and on the alignment of the JPEG grid. Recently, a deep learning-based estimator capable to work under a wide variety of conditions has been proposed, that outperforms tailored existing methods in most of the cases. The method is based on a Convolutional Neural Network (CNN) that is trained to solve the estimation as a standard regression problem. By exploiting the integer nature of the quantization coefficients, in this paper, we propose a deep learning technique that performs the estimation by resorting to a simil-classification architecture. The CNN is trained with a loss function that takes into account both the accuracy and the Mean Square Error (MSE) of the estimation. Results confirm the superior performance of the proposed technique, compared to the state-of-the art methods based on statistical analysis and, in particular, deep learning regression. Moreover, the capability of the method to work under general operative conditions, regarding the alignment of the second compression grid with the one of first compression and the combinations of the JPEG qualities of former and second compression, is very relevant in practical applications, where these information are unknown a priori.

Xingran Zhou, Siyu Huang, Bin Li et al.

This paper presents a novel method to manipulate the visual appearance (pose and attribute) of a person image according to natural language descriptions. Our method can be boiled down to two stages: 1) text guided pose generation and 2) visual appearance transferred image synthesis. In the first stage, our method infers a reasonable target human pose based on the text. In the second stage, our method synthesizes a realistic and appearance transferred person image according to the text in conjunction with the target pose. Our method extracts sufficient information from the text and establishes a mapping between the image space and the language space, making generating and editing images corresponding to the description possible. We conduct extensive experiments to reveal the effectiveness of our method, as well as using the VQA Perceptual Score as a metric for evaluating the method. It shows for the first time that we can automatically edit the person image from the natural language descriptions.

Yiding Liu, Siyu Yang, Bin Li et al.

We present an instance segmentation scheme based on pixel affinity information, which is the relationship of two pixels belonging to a same instance. In our scheme, we use two neural networks with similar structure. One is to predict pixel level semantic score and the other is designed to derive pixel affinities. Regarding pixels as the vertexes and affinities as edges, we then propose a simple yet effective graph merge algorithm to cluster pixels into instances. Experimental results show that our scheme can generate fine-grained instance mask. With Cityscapes training data, the proposed scheme achieves 27.3 AP on test set.

Li Li, Zhu Li, Bin Li et al.

In this paper, we present a novel pseudo sequence based 2-D hierarchical reference structure for light-field image compression. In the proposed scheme, we first decompose the light-field image into multiple views and organize them into a 2-D coding structure according to the spatial coordinates of the corresponding microlens. Then we mainly develop three technologies to optimize the 2-D coding structure. First, we divide all the views into four quadrants, and all the views are encoded one quadrant after another to reduce the reference buffer size as much as possible. Inside each quadrant, all the views are encoded hierarchically to fully exploit the correlations between different views. Second, we propose to use the distance between the current view and its reference views as the criteria for selecting better reference frames for each inter view. Third, we propose to use the spatial relative positions between different views to achieve more accurate motion vector scaling. The whole scheme is implemented in the reference software of High Efficiency Video Coding. The experimental results demonstrate that the proposed novel pseudo-sequence based 2-D hierarchical structure can achieve maximum 14.2% bit-rate savings compared with the state-of-the-art light-field image compression method.

Jiahao Li, Bin Li, Jizheng Xu et al.

Traditional intra prediction usually utilizes the nearest reference line to generate the predicted block when considering strong spatial correlation. However, this kind of single line-based method does not always work well due to at least two issues. One is the incoherence caused by the signal noise or the texture of other object, where this texture deviates from the inherent texture of the current block. The other reason is that the nearest reference line usually has worse reconstruction quality in block-based video coding. Due to these two issues, this paper proposes an efficient multiple line-based intra prediction scheme to improve coding efficiency. Besides the nearest reference line, further reference lines are also utilized. The further reference lines with relatively higher quality can provide potential better prediction. At the same time, the residue compensation is introduced to calibrate the prediction of boundary regions in a block when we utilize further reference lines. To speed up the encoding process, this paper designs several fast algorithms. Experimental results show that, compared with HM-16.9, the proposed fast search method achieves 2.0% bit saving on average and up to 3.7%, with increasing the encoding time by 112%.