Most existing Low-light Image Enhancement (LLIE) methods either directly map Low-Light (LL) to Normal-Light (NL) images or use semantic or illumination maps as guides. However, the ill-posed nature of LLIE and the difficulty of semantic retrieval from impaired inputs limit these methods, especially in extremely low-light conditions. To address this issue, we present a new LLIE network via Generative LAtent feature based codebook REtrieval (GLARE), in which the codebook prior is derived from undegraded NL images using a Vector Quantization (VQ) strategy. More importantly, we develop a generative Invertible Latent Normalizing Flow (I-LNF) module to align the LL feature distribution to NL latent representations, guaranteeing the correct code retrieval in the codebook. In addition, a novel Adaptive Feature Transformation (AFT) module, featuring an adjustable function for users and comprising an Adaptive Mix-up Block (AMB) along with a dual-decoder architecture, is devised to further enhance fidelity while preserving the realistic details provided by codebook prior. Extensive experiments confirm the superior performance of GLARE on various benchmark datasets and real-world data. Its effectiveness as a preprocessing tool in low-light object detection tasks further validates GLARE for high-level vision applications. Code is released at https://github.com/LowLevelAI/GLARE.
In FDD massive multiple-input multiple-output (MIMO) systems, user equipments (UEs) are required to send back the downlink channel state information (CSI) to base stations (BSs). However, the high-dimentional CSI souring proportional to antennas leads to unacceptable feedback overhead. Recently, deep learning (DL) has shown its unique powerful ability to compress and reconstruct CSI. In this paper, a novel Multilayer Perceptron (MLP) based feedback network, named M-Net, is proposed, where the specific spatial correlation of channels is emphasized by tailored feature processing modules. To meet the practical requirements of lightweight networks and ensure low inference latency, M-Net adopts parameter sharing to reduce network complexity and eliminates complex operations. Additionally, to leverage the computational resources at the BSs, we design a scalable decoder to further improve the performance of the network. Simulation results show that the proposed M-Net achieves almost the state-of-the-art (SOTA) performance with the lowest computational complexity, while exhibiting a remarkable 49.63% reduction in the inference delay compared to TransNet. Finally, we demonstrate that the M-Net variants achieve the SOTA performance by only deep enning the M-Net decoder. The open source codes are available at https://github.com/Password-YYX/M-Net.
Frequency-modulated continuous-wave (FMCW) radar plays a pivotal role in the field of remote sensing. The increasing degree of FMCW radar deployment has increased the mutual interference, which weakens the detection capabilities of radars and threatens reliability and safety of systems. In this paper, a novel FMCW radar interference mitigation (RIM) method, termed as RIMformer, is proposed by using an end-to-end Transformer-based structure. In the RIMformer, a dual multi-head self-attention mechanism is proposed to capture the correlations among the distinct distance elements of intermediate frequency (IF) signals. Additionally, an improved convolutional block is integrated to harness the power of convolution for extracting local features. The architecture is designed to process time-domain IF signals in an end-to-end manner, thereby avoiding the need for additional manual data processing steps. The improved decoder structure ensures the parallelization of the network to increase its computational efficiency. Simulation and measurement experiments are carried out to validate the accuracy and effectiveness of the proposed method. The results show that the proposed RIMformer can effectively mitigate interference and restore the target signals.
Distributed computing in the context of deep neural networks (DNNs) implies the execution of one part of the network on edge devices and the other part typically on a large-scale cloud platform. Conventional methods propose to employ a serial concatenation of a learned image and source encoder, the latter projecting the image encoder output (bottleneck features) into a quantized representation for bitrate-efficient transmission. In the cloud, a respective source decoder reprojects the quantized representation to the original feature representation, serving as an input for the downstream task decoder performing, e.g., semantic segmentation. In this work, we propose joint source and task decoding, as it allows for a smaller network size in the cloud. This further enables the scalability of such services in large numbers without requiring extensive computational load on the cloud per channel. We demonstrate the effectiveness of our method by achieving a distributed semantic segmentation SOTA over a wide range of bitrates on the mean intersection over union metric, while using only $9.8 \%$ ... $11.59 \%$ of cloud DNN parameters used in the previous SOTA on the COCO and Cityscapes datasets.
In quantised autoencoders, images are usually split into local patches, each encoded by one token. This representation is redundant in the sense that the same number of tokens is spend per region, regardless of the visual information content in that region. Adaptive discretisation schemes like quadtrees are applied to allocate tokens for patches with varying sizes, but this just varies the region of influence for a token which nevertheless remains a local descriptor. Modern architectures add an attention mechanism to the autoencoder which infuses some degree of global information into the local tokens. Despite the global context, tokens are still associated with a local image region. In contrast, our method is inspired by spectral decompositions which transform an input signal into a superposition of global frequencies. Taking the data-driven perspective, we learn custom basis functions corresponding to the codebook entries in our VQ-VAE setup. Furthermore, a decoder combines these basis functions in a non-linear fashion, going beyond the simple linear superposition of spectral decompositions. We can achieve this global description with an efficient transpose operation between features and channels and demonstrate our performance on compression.
Sign language translation (SLT) addresses the problem of translating information from a sign language in video to a spoken language in text. Existing studies, while showing progress, are often limited to narrow domains and/or few sign languages and struggle with open-domain tasks. In this paper, we push forward the frontier of SLT by scaling pretraining data, model size, and number of translation directions. We perform large-scale SLT pretraining on different data including 1) noisy multilingual YouTube SLT data, 2) parallel text corpora, and 3) SLT data augmented by translating video captions to other languages with off-the-shelf machine translation models. We unify different pretraining tasks with task-specific prompts under the encoder-decoder architecture, and initialize the SLT model with pretrained (m/By)T5 models across model sizes. SLT pretraining results on How2Sign and FLEURS-ASL#0 (ASL to 42 spoken languages) demonstrate the significance of data/model scaling and cross-lingual cross-modal transfer, as well as the feasibility of zero-shot SLT. We finetune the pretrained SLT models on 5 downstream open-domain SLT benchmarks covering 5 sign languages. Experiments show substantial quality improvements over the vanilla baselines, surpassing the previous state-of-the-art (SOTA) by wide margins.
Since LIC has made rapid progress recently compared to traditional methods, this paper attempts to discuss the question about 'Where is the boundary of Learned Image Compression(LIC)?'. Thus this paper splits the above problem into two sub-problems:1)Where is the boundary of rate-distortion performance of PSNR? 2)How to further improve the compression gain and achieve the boundary? Therefore this paper analyzes the effectiveness of scaling parameters for encoder, decoder and context model, which are the three components of LIC. Then we conclude that scaling for LIC is to scale for context model and decoder within LIC. Extensive experiments demonstrate that overfitting can actually serve as an effective context. By optimizing the context, this paper further improves PSNR and achieves state-of-the-art performance, showing a performance gain of 14.39% with BD-RATE over VVC.
Visual text generation has significantly advanced through diffusion models aimed at producing images with readable and realistic text. Recent works primarily use a ControlNet-based framework, employing standard font text images to control diffusion models. Recognizing the critical role of control information in generating high-quality text, we investigate its influence from three perspectives: input encoding, role at different stages, and output features. Our findings reveal that: 1) Input control information has unique characteristics compared to conventional inputs like Canny edges and depth maps. 2) Control information plays distinct roles at different stages of the denoising process. 3) Output control features significantly differ from the base and skip features of the U-Net decoder in the frequency domain. Based on these insights, we propose TextGen, a novel framework designed to enhance generation quality by optimizing control information. We improve input and output features using Fourier analysis to emphasize relevant information and reduce noise. Additionally, we employ a two-stage generation framework to align the different roles of control information at different stages. Furthermore, we introduce an effective and lightweight dataset for training. Our method achieves state-of-the-art performance in both Chinese and English text generation. The code and dataset available at https://github.com/CyrilSterling/TextGen.
Recently, Handwritten Mathematical Expression Recognition (HMER) has gained considerable attention in pattern recognition for its diverse applications in document understanding. Current methods typically approach HMER as an image-to-sequence generation task within an autoregressive (AR) encoder-decoder framework. However, these approaches suffer from several drawbacks: 1) a lack of overall language context, limiting information utilization beyond the current decoding step; 2) error accumulation during AR decoding; and 3) slow decoding speed. To tackle these problems, this paper makes the first attempt to build a novel bottom-up Non-AutoRegressive Modeling approach for HMER, called NAMER. NAMER comprises a Visual Aware Tokenizer (VAT) and a Parallel Graph Decoder (PGD). Initially, the VAT tokenizes visible symbols and local relations at a coarse level. Subsequently, the PGD refines all tokens and establishes connectivities in parallel, leveraging comprehensive visual and linguistic contexts. Experiments on CROHME 2014/2016/2019 and HME100K datasets demonstrate that NAMER not only outperforms the current state-of-the-art (SOTA) methods on ExpRate by 1.93%/2.35%/1.49%/0.62%, but also achieves significant speedups of 13.7x and 6.7x faster in decoding time and overall FPS, proving the effectiveness and efficiency of NAMER.
Audio-Visual Segmentation (AVS) aims to achieve pixel-level localization of sound sources in videos, while Audio-Visual Semantic Segmentation (AVSS), as an extension of AVS, further pursues semantic understanding of audio-visual scenes. However, since the AVSS task requires the establishment of audio-visual correspondence and semantic understanding simultaneously, we observe that previous methods have struggled to handle this mashup of objectives in end-to-end training, resulting in insufficient learning and sub-optimization. Therefore, we propose a two-stage training strategy called \textit{Stepping Stones}, which decomposes the AVSS task into two simple subtasks from localization to semantic understanding, which are fully optimized in each stage to achieve step-by-step global optimization. This training strategy has also proved its generalization and effectiveness on existing methods. To further improve the performance of AVS tasks, we propose a novel framework Adaptive Audio Visual Segmentation, in which we incorporate an adaptive audio query generator and integrate masked attention into the transformer decoder, facilitating the adaptive fusion of visual and audio features. Extensive experiments demonstrate that our methods achieve state-of-the-art results on all three AVS benchmarks. The project homepage can be accessed at https://gewu-lab.github.io/stepping_stones/.
The utilization of Transformer-based models prospers the growth of multi-document summarization (MDS). Given the huge impact and widespread adoption of Transformer-based models in various natural language processing tasks, investigating their performance and behaviors in the context of MDS becomes crucial for advancing the field and enhancing the quality of summary. To thoroughly examine the behaviours of Transformer-based MDS models, this paper presents five empirical studies on (1) measuring the impact of document boundary separators quantitatively; (2) exploring the effectiveness of different mainstream Transformer structures; (3) examining the sensitivity of the encoder and decoder; (4) discussing different training strategies; and (5) discovering the repetition in a summary generation. The experimental results on prevalent MDS datasets and eleven evaluation metrics show the influence of document boundary separators, the granularity of different level features and different model training strategies. The results also reveal that the decoder exhibits greater sensitivity to noises compared to the encoder. This underscores the important role played by the decoder, suggesting a potential direction for future research in MDS. Furthermore, the experimental results indicate that the repetition problem in the generated summaries has correlations with the high uncertainty scores.
In recent years, transformer-based models have exhibited considerable potential in point cloud instance segmentation. Despite the promising performance achieved by existing methods, they encounter challenges such as instance query initialization problems and excessive reliance on stacked layers, rendering them incompatible with large-scale 3D scenes. This paper introduces a novel method, named SGIFormer, for 3D instance segmentation, which is composed of the Semantic-guided Mix Query (SMQ) initialization and the Geometric-enhanced Interleaving Transformer (GIT) decoder. Specifically, the principle of our SMQ initialization scheme is to leverage the predicted voxel-wise semantic information to implicitly generate the scene-aware query, yielding adequate scene prior and compensating for the learnable query set. Subsequently, we feed the formed overall query into our GIT decoder to alternately refine instance query and global scene features for further capturing fine-grained information and reducing complex design intricacies simultaneously. To emphasize geometric property, we consider bias estimation as an auxiliary task and progressively integrate shifted point coordinates embedding to reinforce instance localization. SGIFormer attains state-of-the-art performance on ScanNet V2, ScanNet200 datasets, and the challenging high-fidelity ScanNet++ benchmark, striking a balance between accuracy and efficiency. The code, weights, and demo videos are publicly available at https://rayyoh.github.io/sgiformer.
Existing Video Object Segmentation (VOS) relies on explicit user instructions, such as categories, masks, or short phrases, restricting their ability to perform complex video segmentation requiring reasoning with world knowledge. In this paper, we introduce a new task, Reasoning Video Object Segmentation (ReasonVOS). This task aims to generate a sequence of segmentation masks in response to implicit text queries that require complex reasoning abilities based on world knowledge and video contexts, which is crucial for structured environment understanding and object-centric interactions, pivotal in the development of embodied AI. To tackle ReasonVOS, we introduce VISA (Video-based large language Instructed Segmentation Assistant), to leverage the world knowledge reasoning capabilities of multi-modal LLMs while possessing the ability to segment and track objects in videos with a mask decoder. Moreover, we establish a comprehensive benchmark consisting of 35,074 instruction-mask sequence pairs from 1,042 diverse videos, which incorporates complex world knowledge reasoning into segmentation tasks for instruction-tuning and evaluation purposes of ReasonVOS models. Experiments conducted on 8 datasets demonstrate the effectiveness of VISA in tackling complex reasoning segmentation and vanilla referring segmentation in both video and image domains. The code and dataset are available at https://github.com/cilinyan/VISA.
Image captioning is a technology that produces text-based descriptions for an image. Deep learning-based solutions built on top of feature recognition may very well serve the purpose. But as with any other machine learning solution, the user understanding in the process of caption generation is poor and the model does not provide any explanation for its predictions and hence the conventional methods are also referred to as Black-Box methods. Thus, an approach where the model's predictions are trusted by the user is needed to appreciate interoperability. Explainable AI is an approach where a conventional method is approached in a way that the model or the algorithm's predictions can be explainable and justifiable. Thus, this article tries to approach image captioning using Explainable AI such that the resulting captions generated by the model can be Explained and visualized. A newer architecture with a CNN decoder and hierarchical attention concept has been used to increase speed and accuracy of caption generation. Also, incorporating explainability to a model makes it more trustable when used in an application. The model is trained and evaluated using MSCOCO dataset and both quantitative and qualitative results are presented in this article.
Index modulation (IM) is one of the key enabling technologies for beyond fifth generation (B5G) and sixth generation (6G) wireless systems, attracting attention for its inherent energy and spectral efficiency resulting from conveying information through the indexation of the resources utilized in during signal transmission. However, a remaining critical bottleneck for large-scale IM is the consequently infeasible detection complexity of combinatoric order. Therefore in this article, in order to maximally reap the advantages of IM in large scenarios, we propose a novel message passing (MP) decoder designed under the Gaussian belief propagation (GaBP) framework exploiting a novel unit vector decomposition (UVD) of IM signals with purpose-derived novel probability distributions. The proposed method enjoys a low decoding complexity that is independent of previously prohibitive combinatorial factors, while still approach-ing the performance of unfeasible state-of-the-art (SotA) search-based methods. The effectiveness of the proposed approach is demonstrated via complexity analysis and numerical results for the exemplary piloted generalized quadrature spatial modulation (GQSM) systems of truly massive sizes (up to 96 antennas).
Multi-talker speech recognition and target-talker speech recognition, both involve transcription in multi-talker contexts, remain significant challenges. However, existing methods rarely attempt to simultaneously address both tasks. In this study, we propose a pioneering approach to empower Whisper, which is a speech foundation model, to tackle joint multi-talker and target-talker speech recognition tasks. Specifically, (i) we freeze Whisper and plug a Sidecar separator into its encoder to separate mixed embedding for multiple talkers; (ii) a Target Talker Identifier is introduced to identify the embedding flow of the target talker on the fly, requiring only three-second enrollment speech as a cue; (iii) soft prompt tuning for decoder is explored for better task adaptation. Our method outperforms previous methods on two- and three-talker LibriMix and LibriSpeechMix datasets for both tasks, and delivers acceptable zero-shot performance on multi-talker ASR on AishellMix Mandarin dataset.
Most currently deployed large language models (LLMs) undergo continuous training or additional finetuning. By contrast, most research into LLMs' internal mechanisms focuses on models at one snapshot in time (the end of pre-training), raising the question of whether their results generalize to real-world settings. Existing studies of mechanisms over time focus on encoder-only or toy models, which differ significantly from most deployed models. In this study, we track how model mechanisms, operationalized as circuits, emerge and evolve across 300 billion tokens of training in decoder-only LLMs, in models ranging from 70 million to 2.8 billion parameters. We find that task abilities and the functional components that support them emerge consistently at similar token counts across scale. Moreover, although such components may be implemented by different attention heads over time, the overarching algorithm that they implement remains. Surprisingly, both these algorithms and the types of components involved therein can replicate across model scale. These results suggest that circuit analyses conducted on small models at the end of pre-training can provide insights that still apply after additional pre-training and over model scale.
We present two novel domain-independent genetic operators that harness the capabilities of deep learning: a crossover operator for genetic algorithms and a mutation operator for genetic programming. Deep Neural Crossover leverages the capabilities of deep reinforcement learning and an encoder-decoder architecture to select offspring genes. BERT mutation masks multiple gp-tree nodes and then tries to replace these masks with nodes that will most likely improve the individual's fitness. We show the efficacy of both operators through experimentation.
Scanpath prediction in 360{\deg} images can help realize rapid rendering and better user interaction in Virtual/Augmented Reality applications. However, existing scanpath prediction models for 360{\deg} images execute scanpath prediction on 2D equirectangular projection plane, which always result in big computation error owing to the 2D plane's distortion and coordinate discontinuity. In this work, we perform scanpath prediction for 360{\deg} images in 3D spherical coordinate system and proposed a novel 3D scanpath Transformer named Pathformer3D. Specifically, a 3D Transformer encoder is first used to extract 3D contextual feature representation for the 360{\deg} image. Then, the contextual feature representation and historical fixation information are input into a Transformer decoder to output current time step's fixation embedding, where the self-attention module is used to imitate the visual working memory mechanism of human visual system and directly model the time dependencies among the fixations. Finally, a 3D Gaussian distribution is learned from each fixation embedding, from which the fixation position can be sampled. Evaluation on four panoramic eye-tracking datasets demonstrates that Pathformer3D outperforms the current state-of-the-art methods. Code is available at https://github.com/lsztzp/Pathformer3D .
Multi-organ segmentation in medical image analysis is crucial for diagnosis and treatment planning. However, many factors complicate the task, including variability in different target categories and interference from complex backgrounds. In this paper, we utilize the knowledge of Deformable Convolution V3 (DCNv3) and multi-object segmentation to optimize our Spatially Adaptive Convolution Network (SACNet) in three aspects: feature extraction, model architecture, and loss constraint, simultaneously enhancing the perception of different segmentation targets. Firstly, we propose the Adaptive Receptive Field Module (ARFM), which combines DCNv3 with a series of customized block-level and architecture-level designs similar to transformers. This module can capture the unique features of different organs by adaptively adjusting the receptive field according to various targets. Secondly, we utilize ARFM as building blocks to construct the encoder-decoder of SACNet and partially share parameters between the encoder and decoder, making the network wider rather than deeper. This design achieves a shared lightweight decoder and a more parameter-efficient and effective framework. Lastly, we propose a novel continuity dynamic adjustment loss function, based on t-vMF dice loss and cross-entropy loss, to better balance easy and complex classes in segmentation. Experiments on 3D slice datasets from ACDC and Synapse demonstrate that SACNet delivers superior segmentation performance in multi-organ segmentation tasks compared to several existing methods.