Bridging different modalities lies at the heart of cross-modality generation. While conventional approaches treat the text modality as a conditioning signal that gradually guides the denoising process from Gaussian noise to the target image modality, we explore a much simpler paradigm-directly evolving between text and image modalities through flow matching. This requires projecting both modalities into a shared latent space, which poses a significant challenge due to their inherently different representations: text is highly semantic and encoded as 1D tokens, whereas images are spatially redundant and represented as 2D latent embeddings. To address this, we introduce FlowTok, a minimal framework that seamlessly flows across text and images by encoding images into a compact 1D token representation. Compared to prior methods, this design reduces the latent space size by 3.3x at an image resolution of 256, eliminating the need for complex conditioning mechanisms or noise scheduling. Moreover, FlowTok naturally extends to image-to-text generation under the same formulation. With its streamlined architecture centered around compact 1D tokens, FlowTok is highly memory-efficient, requires significantly fewer training resources, and achieves much faster sampling speeds-all while delivering performance comparable to state-of-the-art models. Code is available at https://github.com/TACJu/FlowTok.

Image tokenizers form the foundation of modern text-to-image generative models but are notoriously difficult to train. Furthermore, most existing text-to-image models rely on large-scale, high-quality private datasets, making them challenging to replicate. In this work, we introduce Text-Aware Transformer-based 1-Dimensional Tokenizer (TA-TiTok), an efficient and powerful image tokenizer that can utilize either discrete or continuous 1-dimensional tokens. TA-TiTok uniquely integrates textual information during the tokenizer decoding stage (i.e., de-tokenization), accelerating convergence and enhancing performance. TA-TiTok also benefits from a simplified, yet effective, one-stage training process, eliminating the need for the complex two-stage distillation used in previous 1-dimensional tokenizers. This design allows for seamless scalability to large datasets. Building on this, we introduce a family of text-to-image Masked Generative Models (MaskGen), trained exclusively on open data while achieving comparable performance to models trained on private data. We aim to release both the efficient, strong TA-TiTok tokenizers and the open-data, open-weight MaskGen models to promote broader access and democratize the field of text-to-image masked generative models.

Autoregressive (AR) modeling, known for its next-token prediction paradigm, underpins state-of-the-art language and visual generative models. Traditionally, a ``token'' is treated as the smallest prediction unit, often a discrete symbol in language or a quantized patch in vision. However, the optimal token definition for 2D image structures remains an open question. Moreover, AR models suffer from exposure bias, where teacher forcing during training leads to error accumulation at inference. In this paper, we propose xAR, a generalized AR framework that extends the notion of a token to an entity X, which can represent an individual patch token, a cell (a $k\times k$ grouping of neighboring patches), a subsample (a non-local grouping of distant patches), a scale (coarse-to-fine resolution), or even a whole image. Additionally, we reformulate discrete token classification as continuous entity regression, leveraging flow-matching methods at each AR step. This approach conditions training on noisy entities instead of ground truth tokens, leading to Noisy Context Learning, which effectively alleviates exposure bias. As a result, xAR offers two key advantages: (1) it enables flexible prediction units that capture different contextual granularity and spatial structures, and (2) it mitigates exposure bias by avoiding reliance on teacher forcing. On ImageNet-256 generation benchmark, our base model, xAR-B (172M), outperforms DiT-XL/SiT-XL (675M) while achieving 20$\times$ faster inference. Meanwhile, xAR-H sets a new state-of-the-art with an FID of 1.24, running 2.2$\times$ faster than the previous best-performing model without relying on vision foundation modules (e.g., DINOv2) or advanced guidance interval sampling.

This paper presents Randomized AutoRegressive modeling (RAR) for visual generation, which sets a new state-of-the-art performance on the image generation task while maintaining full compatibility with language modeling frameworks. The proposed RAR is simple: during a standard autoregressive training process with a next-token prediction objective, the input sequence-typically ordered in raster form-is randomly permuted into different factorization orders with a probability r, where r starts at 1 and linearly decays to 0 over the course of training. This annealing training strategy enables the model to learn to maximize the expected likelihood over all factorization orders and thus effectively improve the model's capability of modeling bidirectional contexts. Importantly, RAR preserves the integrity of the autoregressive modeling framework, ensuring full compatibility with language modeling while significantly improving performance in image generation. On the ImageNet-256 benchmark, RAR achieves an FID score of 1.48, not only surpassing prior state-of-the-art autoregressive image generators but also outperforming leading diffusion-based and masked transformer-based methods. Code and models will be made available at https://github.com/bytedance/1d-tokenizer

Autoregressive (AR) modeling has achieved remarkable success in natural language processing by enabling models to generate text with coherence and contextual understanding through next token prediction. Recently, in image generation, VAR proposes scale-wise autoregressive modeling, which extends the next token prediction to the next scale prediction, preserving the 2D structure of images. However, VAR encounters two primary challenges: (1) its complex and rigid scale design limits generalization in next scale prediction, and (2) the generator’s dependence on a discrete tokenizer with the same complex scale structure restricts modularity and flexibility in updating the tokenizer. To address these limitations, we introduce FlowAR, a general next scale prediction method featuring a streamlined scale design, where each subsequent scale is simply double the previous one. This eliminates the need for VAR’s intricate multi-scale residual tokenizer and enables the use of any off-the-shelf Variational AutoEncoder (VAE). Our simplified design enhances generalization in next scale prediction and facilitates the integration of Flow Matching for high-quality image synthesis. We validate the effectiveness of FlowAR on the challenging ImageNet-256 benchmark, demonstrating superior generation performance compared to previous methods. Codes is available at \href{https://github.com/OliverRensu/FlowAR}{https://github.com/OliverRensu/FlowAR}.

While significant advancements have been made in compressed representations for text embeddings in large language models (LLMs), the compression of visual tokens in multi-modal LLMs (MLLMs) has remained a largely overlooked area. In this work, we present the study on the analysis of redundancy concerning visual tokens and efficient training within these models. Our initial experiments show that eliminating up to 70% of visual tokens at the testing stage by simply average pooling only leads to a minimal 3% reduction in visual question answering accuracy on the GQA benchmark, indicating significant redundancy in visual context. Addressing this, we introduce Visual Context Compressor, which reduces the number of visual tokens to enhance training and inference efficiency without sacrificing performance. To minimize information loss caused by the compression on visual tokens while maintaining training efficiency, we develop LLaVolta as a light and staged training scheme that incorporates stage-wise visual context compression to progressively compress the visual tokens from heavily to lightly compression during training, yielding no loss of information when testing. Extensive experiments demonstrate that our approach enhances the performance of MLLMs in both image-language and video-language understanding, while also significantly cutting training costs and improving inference efficiency.

This paper presents innovative enhancements to diffusion models by integrating a novel multi-resolution network and time-dependent layer normalization. Diffusion models have gained prominence for their effectiveness in high-fidelity image generation. While conventional approaches rely on convolutional U-Net architectures, recent Transformer-based designs have demonstrated superior performance and scalability. However, Transformer architectures, which tokenize input data (via "patchification"), face a trade-off between visual fidelity and computational complexity due to the quadratic nature of self-attention operations concerning token length. While larger patch sizes enable attention computation efficiency, they struggle to capture fine-grained visual details, leading to image distortions. To address this challenge, we propose augmenting the **Di**ffusion model with the **M**ulti-**R**esolution network (DiMR), a framework that refines features across multiple resolutions, progressively enhancing detail from low to high resolution. Additionally, we introduce Time-Dependent Layer Normalization (TD-LN), a parameter-efficient approach that incorporates time-dependent parameters into layer normalization to inject time information and achieve superior performance. Our method's efficacy is demonstrated on the class-conditional ImageNet generation benchmark, where DiMR-XL variants surpass previous diffusion models, achieving FID scores of 1.70 on ImageNet $256 \times 256$ and 2.89 on ImageNet $512 \times 512$. Our best variant, DiMR-G, further establishes a state-of-the-art 1.63 FID on ImageNet $256 \times 256$.

Forward-looking sonar (FLS) image segmentation plays a significant role in ocean engineering. However, the existing image segmentation algorithms present difficulties in extracting features from FLS images with weak semantic information, complex backgrounds and strong environmental noises. Convolutional neural networks (CNNs) have demonstrated remarkable capabilities in semantic segmentation tasks, but the locality of convolution limits the ability to extract global context and long-range semantic information. The effective extraction of global contextual information is indispensable for achieving accurate segmentation results in sonar image processing. In this paper, we propose a novel semantic segmentation architecture for forward-looking sonar images called SonarNet. SonarNet is based on a hybrid CNN-Transformer-HOG framework and comprises four modules. 1) The Global-Local Encoder can extract both global and detailed feature information of the underwater target; 2) the Network Decoder converts the high-semantic feature map into a pixel-level classification; 3) as a bridge between dual encoders, the Global-Local Fusion Module ensures semantic consistency between different encoders; 4) the HOG Feature Encoder and Fusion can extract traditional manual features and perform feature alignment. We conducted comprehensive ablation experiments to validate the efficacy of the designed modules. Finally, experimentation revealed that SonarNet significantly outperforms other CNN-based and CNN-Transformer FLS image segmentation methods.

While the intricate underwater environment leads to blurry and faint features of sonar targets, SonarNet has depicted great success due to its high model capabilities and multifeature fusion mechanism. However, their remarkable performance is accompanied by heavier backbones and larger model sizes to achieve benefits at the cost of increased complexity. The research on fast segmenters deployed on edge devices is urgently inquired. In this article, we analyze the current best-performing sonar image segmentation network named SonarNet. Based on the analysis, we propose a lightweight local feature grafted vision transformer (ViT) embedding network for forward-looking sonar (FLS) images called EsonarNet, which promotes a priority balance between efficiency and accuracy. EsonarNet is based on a hybrid local-global feature grafting architecture and comprises four modules. First, expand from the traditional convolutional neural network (CNN) and histogram of oriented gradients (HOG), a lightweight sonar semantic segmentation model based on hybrid CNN-transformer-HOG fusion encoding and decoding, while preserving high efficiency applied to hardware resources. Second, lightweight encoder units are employed in our EsonarNet, including a designed spatial mobile inverted bottleneck convolution (SMBConv) and efficient vision transformer (ViT) module. Third, serving as a transitional liaison between the CNN encoder and the transformer encoder, the local-global features interaction (LGFI) module focuses on dispersing local semantic information to facilitate long-distance computations, and the global-local aggregation unit (GLAU) module computes correlations through dot products to restore inductive bias. Fourth, the HOG features are introduced into EsonarNet through the lightweight HOG-deep learning graft mechanism (LHDGM) module to ensure the coherence and compatibility of the acquired traditional and abstract information with different semantics. Ultimately, experimental results demonstrate that EsonarNet outperforms other methods for FLS image segmentation in efficiency.

In this work, we present a robust approach for joint part and object segmentation. Specifically, we reformulate object and part segmentation as an optimization problem and build a hierarchical feature representation including pixel, part, and object-level embeddings to solve it in a bottom-up clustering manner. Pixels are grouped into several clusters where the part-level embeddings serve as cluster centers. Afterwards, object masks are obtained by compositing the part proposals. This bottom-up interaction is shown to be effective in integrating information from lower semantic levels to higher semantic levels. Based on that, our novel approach Compositor produces part and object segmentation masks simultaneously while improving the mask quality. Compositor achieves state-of-the-art performance on PartImageNet and Pascal-Part by outperforming previous methods by around 0.9% and 1.3% on PartImageNet, 0.4% and 1.7% on Pascal-Part in terms of part and object mIoU and demonstrates better robustness against occlusion by around 4.4% and 7.1% on part and object respectively.

Open-vocabulary segmentation is a challenging task requiring segmenting and recognizing objects from an open set of categories in diverse environments. One way to address this challenge is to leverage multi-modal models, such as CLIP, to provide image and text features in a shared embedding space, which effectively bridges the gap between closed-vocabulary and open-vocabulary recognition. Hence, existing methods often adopt a two-stage framework to tackle the problem, where the inputs first go through a mask generator and then through the CLIP model along with the predicted masks. This process involves extracting features from raw images multiple times, which can be ineffective and inefficient. By contrast, we propose to build everything into a single-stage framework using a _shared **F**rozen **C**onvolutional **CLIP** backbone_, which not only significantly simplifies the current two-stage pipeline, but also remarkably yields a better accuracy-cost trade-off. The resulting single-stage system, called FC-CLIP, benefits from the following observations: the _frozen_ CLIP backbone maintains the ability of open-vocabulary classification and can also serve as a strong mask generator, and the _convolutional_ CLIP generalizes well to a larger input resolution than the one used during contrastive image-text pretraining. Surprisingly, FC-CLIP advances state-of-the-art results on various benchmarks, while running practically fast. Specifically, when training on COCO panoptic data only and testing in a zero-shot manner, FC-CLIP achieve 26.8 PQ, 16.8 AP, and 34.1 mIoU on ADE20K, 18.2 PQ, 27.9 mIoU on Mapillary Vistas, 44.0 PQ, 26.8 AP, 56.2 mIoU on Cityscapes, outperforming the prior art under the same setting by +4.2 PQ, +2.4 AP, +4.2 mIoU on ADE20K, +4.0 PQ on Mapillary Vistas and +20.1 PQ on Cityscapes, respectively. Additionally, the training and testing time of FC-CLIP is 7.5x and 6.6x significantly faster than the same prior art, while using 5.9x fewer total model parameters. Meanwhile, FC-CLIP also sets a new state-of-the-art performance across various open-vocabulary semantic segmentation datasets. Code and models are available at https://github.com/bytedance/fc-clip

Few-shot image classification consists of two consecutive learning processes: 1) In the meta-learning stage, the model acquires a knowledge base from a set of training classes. 2) During meta-testing, the acquired knowledge is used to recognize unseen classes from very few examples. Inspired by the compositional representation of objects in humans, we train a neural network architecture that explicitly represents objects as a dictionary of shared components and their spatial composition. In particular, during meta-learning, we train a knowledge base that consists of a dictionary of component representations and a dictionary of component activation maps that encode common spatial activation patterns of components. The elements of both dictionaries are shared among the training classes. During meta-testing, the representation of unseen classes is learned using the component representations and the component activation maps from the knowledge base. Finally, an attention mechanism is used to strengthen those components that are most important for each category. We demonstrate the value of our interpretable compositional learning framework for a few-shot classification using miniImageNet, tieredImageNet, CIFAR-FS, and FC100, where we achieve comparable performance.

Fine-grained visual classification (FGVC) which aims at recognizing objects from subcategories is a very challenging task due to the inherently subtle inter-class differences. Most existing works mainly tackle this problem by reusing the backbone network to extract features of detected discriminative regions. However, this strategy inevitably complicates the pipeline and pushes the proposed regions to contain most parts of the objects thus fails to locate the really important parts. Recently, vision transformer (ViT) shows its strong performance in the traditional classification task. The self-attention mechanism of the transformer links every patch token to the classification token. In this work, we first evaluate the effectiveness of the ViT framework in the fine-grained recognition setting. Then motivated by the strength of the attention link can be intuitively considered as an indicator of the importance of tokens, we further propose a novel Part Selection Module that can be applied to most of the transformer architectures where we integrate all raw attention weights of the transformer into an attention map for guiding the network to effectively and accurately select discriminative image patches and compute their relations. A contrastive loss is applied to enlarge the distance between feature representations of confusing classes. We name the augmented transformer-based model TransFG and demonstrate the value of it by conducting experiments on five popular fine-grained benchmarks where we achieve state-of-the-art performance. Qualitative results are presented for better understanding of our model.

Mixup-based augmentation has been found to be effective for generalizing models during training, especially for Vision Transformers (ViTs) since they can easily overfit. However, previous mixup-based methods have an underlying prior knowledge that the linearly interpolated ratio of targets should be kept the same as the ratio proposed in input interpolation. This may lead to a strange phenomenon that sometimes there is no valid object in the mixed image due to the random process in augmentation but there is still response in the label space. To bridge such gap between the input and label spaces, we propose TransMix, which mixes labels based on the attention maps of Vision Transformers. The confidence of the label will be larger if the corresponding input image is weighted higher by the attention map. TransMix is embarrassingly simple and can be implemented in just a few lines of code without introducing any extra parameters and FLOPs to ViT-based models. Experimental results show that our method can consistently improve various ViT-based models at scales on ImageNet classification. After pre-trained with TransMix on ImageNet, the ViT-based models also demonstrate better transferability to semantic segmentation, object detection and instance segmentation. TransMix also exhibits to be more robust when evaluating on 4 different benchmarks. Code is publicly available at https://github.com/Beckschen/TransMix.

Semi-supervised video action recognition tends to enable deep neural networks to achieve remarkable performance even with very limited labeled data. However, existing methods are mainly transferred from current image-based methods (e.g., FixMatch). Without specifically utilizing the temporal dynamics and inherent multimodal attributes, their results could be suboptimal. To better leverage the encoded temporal information in videos, we introduce temporal gradient as an additional modality for more attentive feature extraction in this paper. To be specific, our method explicitly distills the fine-grained motion representations from temporal gradient (TG) and imposes consistency across different modalities (i.e., RGB and TG). The performance of semi-supervised action recognition is significantly improved without additional computation or parameters during inference. Our method achieves the state-of-the-art performance on three video action recognition benchmarks (i.e., Kinetics-400, UCF-101, and HMDB-51) under several typical semi-supervised settings (i.e., different ratios of labeled data). Code is made available at https://github.com/lambert-x/video-semisup.

Recent work has shown that deep convolutional neural networks (DCNNs) do not generalize well under partial occlusion. Inspired by the success of compositional models at classifying partially occluded objects, we propose to integrate compositional models and DCNNs into a unified deep model with innate robustness to partial occlusion. We term this architecture Compositional Convolutional Neural Network. In particular, we propose to replace the fully connected classification head of a DCNN with a differentiable compositional model. The generative nature of the compositional model enables it to localize occluders and subsequently focus on the non-occluded parts of the object. We conduct classification experiments on artificially occluded images as well as real images of partially occluded objects from the MS-COCO dataset. The results show that DCNNs do not classify occluded objects robustly, even when trained with data that is strongly augmented with partial occlusions. Our proposed model outperforms standard DCNNs by a large margin at classifying partially occluded objects, even when it has not been exposed to occluded objects during training. Additional experiments demonstrate that CompositionalNets can also localize the occluders accurately, despite being trained with class labels only. The code and data used in this work are publicly available.

With the prevalence of depth sensors and 3D scanning devices, point clouds have attracted increasing attention as a format for 3D object representation, with applications in various fields such as tele-presence, navigation for autonomous driving and heritage reconstruction. However, point clouds usually exhibit holes of missing data, mainly due to the limitation of acquisition techniques and complicated structure. Hence, we propose an efficient inpainting method for the attribute (e.g., color) of point clouds, exploiting non-local self-similarity in graph spectral domain. Specifically, we represent irregular point clouds naturally on graphs, and split a point cloud into fixed-sized cubes as the processing unit. We then globally search for the most similar cubes to the target cube with holes inside, and compute the graph Fourier transform (GFT) basis from the similar cubes, which will be leveraged for the GFT representation of the target patch. We then formulate attribute inpainting as a sparse coding problem, imposing sparsity on the GFT representation of the attribute for hole filling. Experimental results demonstrate the superiority of our method.