Based on pruning techniques that interpret polar codes as low-density parity-check-like (LDPC-like) codes, this paper develops novel belief propagation (BP)-based decoders utilizing the concept of schedule diversity. By analyzing the impact of hidden variable nodes (VNHs), scheduling algorithms that guarantee fast convergence speeds and superior error performance are devised for the LDPC-like sparse factor graph (SPF), incorporating both standard sequential scheduling (SSS) and informed dynamic scheduling (IDS) algorithms. To further enhance error performance, cyclic redundancy check (CRC) bits are integrated into the LDPC-like SPF, and random update sequences are introduced to reduce computational complexity while implementing schedule diversity. Inspired by belief propagation list decoding, the proposed approaches eventually update messages in the LDPC-like SPF according to various schedules, thereby producing diverse decoding outcomes. Compared to the CRC-aided successive cancellation list (CA-SCL) decoder and benchmark soft-output decoders in the literature, the proposed schemes achieve comparable or improved error performance while reducing computational complexity in the high SNR region. These results serve as preliminary findings on the BP algorithm over LDPC-like SFGs and provide useful insights for the design and practical improvement of polar decoders.

Most frame-based learned video codecs can be interpreted as recurrent neural networks (RNNs) propagating reference information along the temporal dimension. This work revisits the limitations of the current approaches from an RNN perspective. The output-recurrence methods, which propagate decoded frames, are intuitive but impose dual constraints on the output decoded frames, leading to suboptimal rate-distortion performance. In contrast, the hidden-to-hidden connection approaches, which propagate latent features within the RNN, offer greater flexibility but require large buffer sizes. To address these issues, we propose HyTIP, a learned video coding framework that combines both mechanisms. Our hybrid buffering strategy uses explicit decoded frames and a small number of implicit latent features to achieve competitive coding performance. Experimental results show that our HyTIP outperforms the sole use of either output-recurrence or hidden-to-hidden approaches. Furthermore, it achieves comparable performance to state-of-the-art methods but with a much smaller buffer size, and outperforms VTM 17.0 (Low-delay B) in terms of PSNR-RGB and MS-SSIM-RGB. The source code of HyTIP is available at https://github.com/NYCU-MAPL/HyTIP.

In this work, we present a learned multi-task video codec that is optimized for human and machine vision. The codec consists of an encoder that maps images from the pixel domain to a latent representation and multiple decoders that map the latent to either an image for human consumption or multiple task-specific features for different machine vision tasks. This allows a single bitstream to be used for multiple tasks while also reducing the decoder complexity for machine vision tasks. Unlike most learned codecs, our method performs inter-coding at the latent level instead of the pixel domain. Experiments show that the proposed method achieves a compression performance for machine vision tasks comparable to other multi-task codecs designed for machine vision only, while also providing video reconstruction. The code is available at https://github.com/GreenAutoML4FAS/HybridMultiTaskCoding.

This paper presents the first-ever study of adapting compressed image latents to suit the needs of downstream vision tasks that adopt Multimodal Large Language Models (MLLMs). MLLMs have extended the success of large language models to modalities (e.g. images) beyond text, but their billion scale hinders deployment on resource-constrained end devices. While cloud-hosted MLLMs could be available, transmitting raw, uncompressed images captured by end devices to the cloud requires an efficient image compression system. To address this, we focus on emerging neural image compression and propose a novel framework with a lightweight transform-neck and a surrogate loss to adapt compressed image latents for MLLM-based vision tasks. Given the huge scale of MLLMs, our framework excludes the entire downstream MLLM except part of its visual encoder from training our system. This stands out from most existing coding for machine approaches that involve downstream networks in training and thus could be impractical when the networks are MLLMs. The proposed framework is general in that it is applicable to various MLLMs, neural image codecs, and multiple application scenarios, where the neural image codec can be (1) pre-trained for human perception without updating, (2) fully updated for joint human and machine perception, or (3) fully updated for only machine perception. Extensive experiments on different neural image codecs and various MLLMs show that our method achieves great rate-accuracy performance with much less complexity.

3D Gaussian Splatting (3DGS) has recently emerged as a promising 3D representation. Much research has been focused on reducing its storage requirements and memory footprint. However, the needs to compress and transmit the 3DGS representation to the remote side are overlooked. This new application calls for rate-distortion-optimized 3DGS compression. How to quantize and entropy encode sparse Gaussian primitives in the 3D space remains largely unexplored. Few early attempts resort to the hyperprior framework from learned image compression. But, they fail to utilize fully the inter and intra correlation inherent in Gaussian primitives. Built on ScaffoldGS, this work, termed CAT-3DGS, introduces a context-adaptive triplane approach to their rate-distortion-optimized coding. It features multi-scale triplanes, oriented according to the principal axes of Gaussian primitives in the 3D space, to capture their inter correlation (i.e. spatial correlation) for spatial autoregressive coding in the projected 2D planes. With these triplanes serving as the hyperprior, we further perform channel-wise autoregressive coding to leverage the intra correlation within each individual Gaussian primitive. Our CAT-3DGS incorporates a view frequency-aware masking mechanism. It actively skips from coding those Gaussian primitives that potentially have little impact on the rendering quality. When trained end-to-end to strike a good rate-distortion trade-off, our CAT-3DGS achieves the state-of-the-art compression performance on the commonly used real-world datasets.

This work proposes a hybrid, explicit-implicit temporal buffering scheme for conditional residual video coding. Recent conditional coding methods propagate implicit temporal information for inter-frame coding, demonstrating superior coding performance to those relying exclusively on previously decoded frames (i.e. the explicit temporal information). However, these methods require substantial memory to store a large number of implicit features. This work presents a hybrid buffering strategy. For inter-frame coding, it buffers one previously decoded frame as the explicit temporal reference and a small number of learned features as implicit temporal reference. Our hybrid buffering scheme for conditional residual coding outperforms the single use of explicit or implicit information. Moreover, it allows the total buffer size to be reduced to the equivalent of two video frames with a negligible performance drop on 2K video sequences. The ablation experiment further sheds light on how these two types of temporal references impact the coding performance.

This paper proposes a learned hierarchical B-frame coding scheme in response to the Grand Challenge on Neural Network-based Video Coding at ISCAS 2025. Recently, masked conditional residual coding emerged as an attractive alternative to the existing inter-frame coding frameworks, including residual coding, conditional coding, and conditional residual coding. In this work, we propose masked conditional residual B-frame coding, termed MaskCRT-B, for YUV420 videos. It features an asymmetric codec architecture that includes one joint YUV encoder and two separate Y and UV decoders. Moreover, it incorporates a bi-directional adaptive fusion module that refines the bi-directional feature maps to better tackle the prediction of the occluded and dis-occluded regions within the input video. MaskCRT-B presents a significant advancement in learned B-frame coding, outperforming the state-of-the-art conditional B-frame codec from the Grand Challenge at ISCAS 2024.

In this work, we propose Progressive Cool-Chic (PCC), a scalable overfitted neural image codec that can decode an image at two different resolutions from a single bitstream. Experiments show that our method reduces the necessary bitrate to encode two representations of the same image by up to 31.54% in terms of BD-rate compared to encoding both representations independently using Cool-Chic while also decreasing the necessary decoding time. The bitstream is structured in a way that the low-resolution image can already be decoded, when only a part of the bitstream has been received. The code is available at https://github.com/mbenjak/progressive-CC.

The incorporation of LiDAR technology into some high-end smartphones has unlocked numerous possibilities across various applications, including photography, image restoration, augmented reality, and more. In this paper, we introduce a novel direction that harnesses LiDAR depth maps to enhance the compression of the corresponding RGB camera images. To the best of our knowledge, this represents the initial exploration in this particular research direction. Specifically, we propose a Transformer-based learned image compression system capable of achieving variable-rate compression using a single model while utilizing the LiDAR depth map as supplementary information for both the encoding and decoding processes. Experimental results demonstrate that integrating LiDAR yields an average PSNR gain of 0.83 dB and an average bitrate reduction of 16% as compared to its absence.

This paper aims to delve into the rate-distortion-complexity trade-offs of modern neural video coding. Recent years have witnessed much research effort being focused on exploring the full potential of neural video coding. Conditional auto encoders have emerged as the mainstream approach to efficient neural video coding. The central theme of conditional auto encoders is to leverage both spatial and temporal information for better conditional coding. However, a recent study indicates that conditional coding may suffer from information bottlenecks, potentially performing worse than traditional residual coding. To address this issue, recent conditional coding methods incorporate a large number of high-resolution features as the condition signal, leading to a considerable increase in the number of multiply-accumulate operations, memory footprint, and model size. Taking DCVC as the common code base, we investigate how the newly proposed conditional residual coding, an emerging new school of thought, and its variants may strike a better balance among rate, distortion, and complexity.

This work introduces a Transformer-based image compression system. It has the flexibility to switch between the standard image reconstruction and the denoising reconstruction from a single compressed bitstream. Instead of training separate decoders for these tasks, we incorporate two add-on modules to adapt a pre-trained image decoder from performing the standard image reconstruction to joint decoding and denoising. Our scheme adopts a two-pronged approach. It features a latent refinement module to refine the latent representation of a noisy input image for reconstructing a noise-free image. Additionally, it incorporates an instance-specific prompt generator that adapts the decoding process to improve on the latent refinement. Experimental results show that our method achieves a similar level of denoising quality to training a separate decoder for joint decoding and denoising at the expense of only a modest increase in the decoder's model size and computational complexity.

Over the past few years, learning-based video compression has become an active research area. However, most works focus on P-frame coding. Learned B-frame coding is under-explored and more challenging. This work introduces a novel B-frame coding framework, termed B-CANF, that exploits conditional augmented normalizing flows for B-frame coding. B-CANF additionally features two novel elements: frame-type adaptive coding and B*-frames. Our frame-type adaptive coding learns better bit allocation for hierarchical B-frame coding by dynamically adapting the feature distributions according to the B-frame type. Our B*-frames allow greater flexibility in specifying the group-of-pictures (GOP) structure by reusing the B-frame codec to mimic P-frame coding, without the need for an additional, separate P-frame codec. On commonly used datasets, B-CANF achieves the state-of-the-art compression performance as compared to the other learned B-frame codecs and shows comparable BD-rate results to HM-16.23 under the random access configuration in terms of PSNR. When evaluated on different GOP structures, our B*-frames achieve similar performance to the additional use of a separate P-frame codec.

Conditional coding has lately emerged as the mainstream approach to learned video compression. However, a recent study shows that it may perform worse than residual coding when the information bottleneck arises. Conditional residual coding was thus proposed, creating a new school of thought to improve on conditional coding. Notably, conditional residual coding relies heavily on the assumption that the residual frame has a lower entropy rate than that of the intra frame. Recognizing that this assumption is not always true due to dis-occlusion phenomena or unreliable motion estimates, we propose a masked conditional residual coding scheme. It learns a soft mask to form a hybrid of conditional coding and conditional residual coding in a pixel adaptive manner. We introduce a Transformer-based conditional autoencoder. Several strategies are investigated with regard to how to condition a Transformer-based autoencoder for inter-frame coding, a topic that is largely under-explored. Additionally, we propose a channel transform module (CTM) to decorrelate the image latents along the channel dimension, with the aim of using the simple hyperprior to approach similar compression performance to the channel-wise autoregressive model. Experimental results confirm the superiority of our masked conditional residual transformer (termed MaskCRT) to both conditional coding and conditional residual coding. On commonly used datasets, MaskCRT shows comparable BD-rate results to VTM-17.0 under the low delay P configuration in terms of PSNR-RGB and outperforms VTM-17.0 in terms of MS-SSIM-RGB. It also opens up a new research direction for advancing learned video compression.

This work addresses continuous space-time video super-resolution (C-STVSR) that aims to up-scale an input video both spatially and temporally by any scaling factors. One key challenge of C-STVSR is to propagate information temporally among the input video frames. To this end, we introduce a space-time local implicit neural function. It has the striking feature of learning forward motion for a continuum of pixels. We motivate the use of forward motion from the perspective of learning individual motion trajectories, as opposed to learning a mixture of motion trajectories with backward motion. To ease motion interpolation, we encode sparsely sampled forward motion extracted from the input video as the contextual input. Along with a reliability-aware splatting and decoding scheme, our framework, termed MoTIF, achieves the state-of-the-art performance on C-STVSR. The source code of MoTIF is available at https://github.com/sichun233746/MoTIF.

This work aims for transferring a Transformer-based image compression codec from human perception to machine perception without fine-tuning the codec. We propose a transferable Transformer-based image compression framework, termed TransTIC. Inspired by visual prompt tuning, TransTIC adopts an instance-specific prompt generator to inject instance-specific prompts to the encoder and task-specific prompts to the decoder. Extensive experiments show that our proposed method is capable of transferring the base codec to various machine tasks and outperforms the competing methods significantly. To our best knowledge, this work is the first attempt to utilize prompting on the low-level image compression task.

This paper proposes a transformer-based learned image compression system. It is capable of achieving variable-rate compression with a single model while supporting the region-of-interest (ROI) functionality. Inspired by prompt tuning, we introduce prompt generation networks to condition the transformer-based autoencoder of compression. Our prompt generation networks generate content-adaptive tokens according to the input image, an ROI mask, and a rate parameter. The separation of the ROI mask and the rate parameter allows an intuitive way to achieve variable-rate and ROI coding simultaneously. Extensive experiments validate the effectiveness of our proposed method and confirm its superiority over the other competing methods.

This paper addresses the problem of class-incremental learning. The model is trained to recognize the classes added incrementally. It thus suffers from the challenging issue of catastrophic forgetting. Stemming from the knowledge distillation idea of attempting to retain the model's knowledge on seen classes while learning the newly-added ones, we advance to further alleviate the catastrophic forgetting via our proposed multi-anchor distillation objective, which is realized by constraining the spatial relationship between the input data and the multiple class embeddings of each seen class in the feature space while training the model. Moreover, since the knowledge distillation for incremental learning generally relies on keeping a replay buffer to store the samples of seen classes, the buffer of limited size brings another issue of class imbalance: the number of samples from each seen class decreases gradually, thus being much smaller than the number of samples from each new class. We therefore propose to introduce the continually-adapted margin into the classification objective for tackling the prediction bias towards new classes caused by the class imbalance. Experiments are conducted on various datasets and settings to demonstrate the effectiveness and superior performance of our proposed techniques in comparison to several state-of-the-art baselines.

In this work, we propose a hybrid learning-based method for layered spatial scalability. Our framework consists of a base layer (BL), which encodes a spatially downsampled representation of the input video using Versatile Video Coding (VVC), and a learning-based enhancement layer (EL), which conditionally encodes the original video signal. The EL is conditioned by two fused prediction signals: a spatial inter-layer prediction signal, that is generated by spatially upsampling the output of the BL using super-resolution, and a temporal inter-frame prediction signal, that is generated by decoder-side motion compensation without signaling any motion vectors. We show that our method outperforms LCEVC and has comparable performance to full-resolution VVC for high-resolution content, while still offering scalability.

This paper explores the potential of a learned two-layer B-frame codec, known as TLZMC. TLZMC is one of the few early attempts that deviate from the hybrid-based coding architecture by skipping motion coding. With TLZMC, a low-resolution base layer is utilized to encode temporally unpredictable information. We address the question of whether adapting the base-layer bitrate can achieve better rate-distortion performance. We apply the feature map modulation technique to enable per-frame bitrate adaptation of the base layer. We then propose and compare three online search strategies for determining the base-layer rate parameter: per-level brute-force search, per-level greedy search, and per-frame greedy search. Experimental results show that our top-performing search strategy achieves 0.6%-15.8% Bjøntegaard-Delta rate savings over TLZMC.

Graph neural networks (GNNs) have been recently applied in natural language processing. Various GNN research studies are proposed to learn node interactions within the local graph of each document that contains words, sentences, or topics for inductive text classification. However, most inductive GNNs that are built on a word graph generally take global word embeddings as node features, without referring to document-wise contextual information. Consequently, we find that BERT models can perform better than inductive GNNs. An intuitive follow-up approach is used to enrich GNNs with contextual embeddings from BERT, yet there is a lack of related research. In this work, we propose a simple yet effective unified model, coined ConTextING, with a joint training mechanism to learn from both document embeddings and contextual word interactions simultaneously. Our experiments show that ConTextING outperforms pure inductive GNNs and BERT-style models. The analyses also highlight the benefits of the sub-word graph and joint training with separated classifiers.

This paper introduces an online motion rate adaptation scheme for learned video compression, with the aim of achieving content-adaptive coding on individual test sequences to mitigate the domain gap between training and test data. It features a patch-level bit allocation map, termed the $\alpha-$map, to trade off between the bit rates for motion and inter-frame coding in a spatially-adaptive manner. We optimize the $\alpha-$map through an online back-propagation scheme at inference time. Moreover, we incorporate a look-ahead mechanism to consider its impact on future frames. Extensive experimental results confirm that the proposed scheme, when integrated into a conditional learned video codec, is able to adapt motion bit rate effectively, showing much improved rate-distortion performance particularly on test sequences with complicated motion characteristics.

This paper addresses the video rescaling task, which arises from the needs of adapting the video spatial resolution to suit individual viewing devices. We aim to jointly optimize video downscaling and upscaling as a combined task. Most recent studies focus on image-based solutions, which do not consider temporal information. We present two joint optimization approaches based on invertible neural networks with coupling layers. Our Long Short-Term Memory Video Rescaling Network (LSTM-VRN) leverages temporal information in the low-resolution video to form an explicit prediction of the missing high-frequency information for upscaling. Our Multi-input Multi-output Video Rescaling Network (MIMO-VRN) proposes a new strategy for downscaling and upscaling a group of video frames simultaneously. Not only do they outperform the image-based invertible model in terms of quantitative and qualitative results, but also show much improved upscaling quality than the video rescaling methods without joint optimization. To our best knowledge, this work is the first attempt at the joint optimization of video downscaling and upscaling.

Despite the recent success of deep-learning based semantic segmentation, deploying a pre-trained road scene segmenter to a city whose images are not presented in the training set would not achieve satisfactory performance due to dataset biases. Instead of collecting a large number of annotated images of each city of interest to train or refine the segmenter, we propose an unsupervised learning approach to adapt road scene segmenters across different cities. By utilizing Google Street View and its time-machine feature, we can collect unannotated images for each road scene at different times, so that the associated static-object priors can be extracted accordingly. By advancing a joint global and class-specific domain adversarial learning framework, adaptation of pre-trained segmenters to that city can be achieved without the need of any user annotation or interaction. We show that our method improves the performance of semantic segmentation in multiple cities across continents, while it performs favorably against state-of-the-art approaches requiring annotated training data.

The processing algorithm described herein processes multi-level scans of mixed documents, those containing continuous tone images as well as text and line art, and produces good quality binary documents which compress well when operated on by run-length based coding schemes such as the CCITT standard MH (Modified Hoffman, ref 1). The binary documents produced feature sharp edges in text areas and an accurate representation of tone, free of contours, in continuous tone areas of the processed documents. In addition, the overall grayness of the original documents can be recovered from the halftoned documents, which permits them to be converted to other representations more appropriate for specific output devices. The simplicity of the algorithm allows it to work in a real time environment and it can be easily implemented on various signal processors and micro-processors.