Yiyi Zhou

Gen Luo, Minglang Huang, Yiyi Zhou et al.

Parameter-efficient transfer learning (PETL) is an emerging research spot aimed at inexpensively adapting large-scale pre-trained models to downstream tasks. Recent advances have achieved great success in saving storage costs for various pre-trained models by updating a small number of parameters instead of full tuning. However, we notice that most existing PETL methods still incur non-negligible latency during inference. In this paper, we propose a parameter-efficient and computational friendly adapter for giant vision models, called RepAdapter. Specifically, we first prove that common adaptation modules can also be seamlessly integrated into most giant vision models via our structural re-parameterization, thereby achieving zero-cost during inference. We then investigate the sparse design and effective placement of adapter structure, helping our RepAdaper obtain other advantages in terms of parameter efficiency and performance. To validate RepAdapter, we conduct extensive experiments on 27 benchmark datasets of three vision tasks, i.e., image and video classifications and semantic segmentation. Experimental results show the superior performance and efficiency of RepAdapter than the state-of-the-art PETL methods. For instance, RepAdapter outperforms full tuning by +7.2% on average and saves up to 25% training time, 20% GPU memory, and 94.6% storage cost of ViT-B/16 on VTAB-1k. The generalization ability of RepAdapter is also well validated by a bunch of vision models. Our source code is released at https://github.com/luogen1996/RepAdapter.

Peng Mi, Li Shen, Tianhe Ren et al.

Deep neural networks often suffer from poor generalization caused by complex and non-convex loss landscapes. One of the popular solutions is Sharpness-Aware Minimization (SAM), which smooths the loss landscape via minimizing the maximized change of training loss when adding a perturbation to the weight. However, we find the indiscriminate perturbation of SAM on all parameters is suboptimal, which also results in excessive computation, i.e., double the overhead of common optimizers like Stochastic Gradient Descent (SGD). In this paper, we propose an efficient and effective training scheme coined as Sparse SAM (SSAM), which achieves sparse perturbation by a binary mask. To obtain the sparse mask, we provide two solutions which are based onFisher information and dynamic sparse training, respectively. In addition, we theoretically prove that SSAM can converge at the same rate as SAM, i.e., $O(\log T/\sqrt{T})$. Sparse SAM not only has the potential for training acceleration but also smooths the loss landscape effectively. Extensive experimental results on CIFAR10, CIFAR100, and ImageNet-1K confirm the superior efficiency of our method to SAM, and the performance is preserved or even better with a perturbation of merely 50% sparsity. Code is availiable at https://github.com/Mi-Peng/Sparse-Sharpness-Aware-Minimization.

Peng Mi, Jianghang Lin, Yiyi Zhou et al.

In this paper, we study teacher-student learning from the perspective of data initialization and propose a novel algorithm called Active Teacher(Source code are available at: \url{https://github.com/HunterJ-Lin/ActiveTeacher}) for semi-supervised object detection (SSOD). Active Teacher extends the teacher-student framework to an iterative version, where the label set is partially initialized and gradually augmented by evaluating three key factors of unlabeled examples, including difficulty, information and diversity. With this design, Active Teacher can maximize the effect of limited label information while improving the quality of pseudo-labels. To validate our approach, we conduct extensive experiments on the MS-COCO benchmark and compare Active Teacher with a set of recently proposed SSOD methods. The experimental results not only validate the superior performance gain of Active Teacher over the compared methods, but also show that it enables the baseline network, ie, Faster-RCNN, to achieve 100% supervised performance with much less label expenditure, ie 40% labeled examples on MS-COCO. More importantly, we believe that the experimental analyses in this paper can provide useful empirical knowledge for data annotation in practical applications.

Weihao Ye, Qiong Wu, Wenhao Lin et al.

Recent progress in Multimodal Large Language Models(MLLMs) often use large image tokens to compensate the visual shortcoming of MLLMs, which not only exhibits obvious redundancy but also greatly exacerbates the already high computation. Token pruning is an effective solution for speeding up MLLMs, but when and how to drop tokens still remains a challenge. In this paper, we propose a novel and training-free approach for the effective visual token pruning of MLLMs, termed FitPrune, which can quickly produce a complete pruning recipe for MLLMs according to a pre-defined budget. Specifically, FitPrune considers token pruning as a statistical problem of MLLM and its objective is to find out an optimal pruning scheme that can minimize the divergence of the attention distributions before and after pruning. In practice, FitPrune can be quickly accomplished based on the attention statistics from a small batch of inference data, avoiding the expensive trials of MLLMs. According to the pruning recipe, an MLLM can directly remove the redundant visual tokens of different examples during inference. To validate FitPrune, we apply it to a set of recent MLLMs, including LLaVA-1.5, LLaVA-HR and LLaVA-NEXT, and conduct extensive experiments on a set of benchmarks. The experimental results show that our FitPrune can not only reduce the computational complexity to a large extent, while retaining high performance, e.g., -54.9% FLOPs for LLaVA-NEXT with only 0.5% accuracy drop. Notably, the pruning recipe can be obtained in about 5 minutes. Our code is available at https://github.com/ywh187/FitPrune.

Gen Luo, Yiyi Zhou, Lei Jin et al.

Semi-supervised object detection (SSOD) is a research hot spot in computer vision, which can greatly reduce the requirement for expensive bounding-box annotations. Despite great success, existing progress mainly focuses on two-stage detection networks like FasterRCNN, while the research on one-stage detectors is often ignored. In this paper, we focus on the semi-supervised learning for the advanced and popular one-stage detection network YOLOv5. Compared with Faster-RCNN, the implementation of YOLOv5 is much more complex, and the various training techniques used in YOLOv5 can also reduce the benefit of SSOD. In addition to this challenge, we also reveal two key issues in one-stage SSOD, which are low-quality pseudo-labeling and multi-task optimization conflict, respectively. To address these issues, we propose a novel teacher-student learning recipe called OneTeacher with two innovative designs, namely Multi-view Pseudo-label Refinement (MPR) and Decoupled Semi-supervised Optimization (DSO). In particular, MPR improves the quality of pseudo-labels via augmented-view refinement and global-view filtering, and DSO handles the joint optimization conflicts via structure tweaks and task-specific pseudo-labeling. In addition, we also carefully revise the implementation of YOLOv5 to maximize the benefits of SSOD, which is also shared with the existing SSOD methods for fair comparison. To validate OneTeacher, we conduct extensive experiments on COCO and Pascal VOC. The extensive experiments show that OneTeacher can not only achieve superior performance than the compared methods, e.g., 15.0% relative AP gains over Unbiased Teacher, but also well handle the key issues in one-stage SSOD. Our source code is available at: https://github.com/luogen1996/OneTeacher.

Peng Mi, Li Shen, Tianhe Ren et al.

Deep neural networks often suffer from poor generalization due to complex and non-convex loss landscapes. Sharpness-Aware Minimization (SAM) is a popular solution that smooths the loss landscape by minimizing the maximized change of training loss when adding a perturbation to the weight. However, indiscriminate perturbation of SAM on all parameters is suboptimal and results in excessive computation, double the overhead of common optimizers like Stochastic Gradient Descent (SGD). In this paper, we propose Sparse SAM (SSAM), an efficient and effective training scheme that achieves sparse perturbation by a binary mask. To obtain the sparse mask, we provide two solutions based on Fisher information and dynamic sparse training, respectively. We investigate the impact of different masks, including unstructured, structured, and $N$:$M$ structured patterns, as well as explicit and implicit forms of implementing sparse perturbation. We theoretically prove that SSAM can converge at the same rate as SAM, i.e., $O(\log T/\sqrt{T})$. Sparse SAM has the potential to accelerate training and smooth the loss landscape effectively. Extensive experimental results on CIFAR and ImageNet-1K confirm that our method is superior to SAM in terms of efficiency, and the performance is preserved or even improved with a perturbation of merely 50\% sparsity. Code is available at https://github.com/Mi-Peng/Systematic-Investigation-of-Sparse-Perturbed-Sharpness-Aware-Minimization-Optimizer.

Yiwei Ma, Jiayi Ji, Xiaoshuai Sun et al.

In this paper, we study the local visual modeling with grid features for image captioning, which is critical for generating accurate and detailed captions. To achieve this target, we propose a Locality-Sensitive Transformer Network (LSTNet) with two novel designs, namely Locality-Sensitive Attention (LSA) and Locality-Sensitive Fusion (LSF). LSA is deployed for the intra-layer interaction in Transformer via modeling the relationship between each grid and its neighbors. It reduces the difficulty of local object recognition during captioning. LSF is used for inter-layer information fusion, which aggregates the information of different encoder layers for cross-layer semantical complementarity. With these two novel designs, the proposed LSTNet can model the local visual information of grid features to improve the captioning quality. To validate LSTNet, we conduct extensive experiments on the competitive MS-COCO benchmark. The experimental results show that LSTNet is not only capable of local visual modeling, but also outperforms a bunch of state-of-the-art captioning models on offline and online testings, i.e., 134.8 CIDEr and 136.3 CIDEr, respectively. Besides, the generalization of LSTNet is also verified on the Flickr8k and Flickr30k datasets

Gen Luo, Yiyi Zhou, Xiaoshuai Sun et al.

Despite the exciting performance, Transformer is criticized for its excessive parameters and computation cost. However, compressing Transformer remains as an open problem due to its internal complexity of the layer designs, i.e., Multi-Head Attention (MHA) and Feed-Forward Network (FFN). To address this issue, we introduce Group-wise Transformation towards a universal yet lightweight Transformer for vision-and-language tasks, termed as LW-Transformer. LW-Transformer applies Group-wise Transformation to reduce both the parameters and computations of Transformer, while also preserving its two main properties, i.e., the efficient attention modeling on diverse subspaces of MHA, and the expanding-scaling feature transformation of FFN. We apply LW-Transformer to a set of Transformer-based networks, and quantitatively measure them on three vision-and-language tasks and six benchmark datasets. Experimental results show that while saving a large number of parameters and computations, LW-Transformer achieves very competitive performance against the original Transformer networks for vision-and-language tasks. To examine the generalization ability, we also apply our optimization strategy to a recently proposed image Transformer called Swin-Transformer for image classification, where the effectiveness can be also confirmed

Haowei Wang, Jiayi Ji, Tianyu Guo et al.

Panoptic Narrative Detection (PND) and Segmentation (PNS) are two challenging tasks that involve identifying and locating multiple targets in an image according to a long narrative description. In this paper, we propose a unified and effective framework called NICE that can jointly learn these two panoptic narrative recognition tasks. Existing visual grounding tasks use a two-branch paradigm, but applying this directly to PND and PNS can result in prediction conflict due to their intrinsic many-to-many alignment property. To address this, we introduce two cascading modules based on the barycenter of the mask, which are Coordinate Guided Aggregation (CGA) and Barycenter Driven Localization (BDL), responsible for segmentation and detection, respectively. By linking PNS and PND in series with the barycenter of segmentation as the anchor, our approach naturally aligns the two tasks and allows them to complement each other for improved performance. Specifically, CGA provides the barycenter as a reference for detection, reducing BDL's reliance on a large number of candidate boxes. BDL leverages its excellent properties to distinguish different instances, which improves the performance of CGA for segmentation. Extensive experiments demonstrate that NICE surpasses all existing methods by a large margin, achieving 4.1% for PND and 2.9% for PNS over the state-of-the-art. These results validate the effectiveness of our proposed collaborative learning strategy. The project of this work is made publicly available at https://github.com/Mr-Neko/NICE.

Gen Luo, Yiyi Zhou, Jiamu Sun et al.

Most of the existing work in one-stage referring expression comprehension (REC) mainly focuses on multi-modal fusion and reasoning, while the influence of other factors in this task lacks in-depth exploration. To fill this gap, we conduct an empirical study in this paper. Concretely, we first build a very simple REC network called SimREC, and ablate 42 candidate designs/settings, which covers the entire process of one-stage REC from network design to model training. Afterwards, we conduct over 100 experimental trials on three benchmark datasets of REC. The extensive experimental results not only show the key factors that affect REC performance in addition to multi-modal fusion, e.g., multi-scale features and data augmentation, but also yield some findings that run counter to conventional understanding. For example, as a vision and language (V&L) task, REC does is less impacted by language prior. In addition, with a proper combination of these findings, we can improve the performance of SimREC by a large margin, e.g., +27.12% on RefCOCO+, which outperforms all existing REC methods. But the most encouraging finding is that with much less training overhead and parameters, SimREC can still achieve better performance than a set of large-scale pre-trained models, e.g., UNITER and VILLA, portraying the special role of REC in existing V&L research.

Tao Chen, Ze Lin, Hui Li et al.

Given the long textual product information and the product image, Multi-modal Product Summarization (MPS) aims to increase customers' desire to purchase by highlighting product characteristics with a short textual summary. Existing MPS methods can produce promising results. Nevertheless, they still 1) lack end-to-end product summarization, 2) lack multi-grained multi-modal modeling, and 3) lack multi-modal attribute modeling. To improve MPS, we propose an end-to-end multi-grained multi-modal attribute-aware product summarization method (MMAPS) for generating high-quality product summaries in e-commerce. MMAPS jointly models product attributes and generates product summaries. We design several multi-grained multi-modal tasks to better guide the multi-modal learning of MMAPS. Furthermore, we model product attributes based on both text and image modalities so that multi-modal product characteristics can be manifested in the generated summaries. Extensive experiments on a real large-scale Chinese e-commence dataset demonstrate that our model outperforms state-of-the-art product summarization methods w.r.t. several summarization metrics. Our code is publicly available at: https://github.com/KDEGroup/MMAPS.

Haowei Wang, Jiayi Ji, Yiyi Zhou et al.

Panoptic Narrative Grounding (PNG) is an emerging cross-modal grounding task, which locates the target regions of an image corresponding to the text description. Existing approaches for PNG are mainly based on a two-stage paradigm, which is computationally expensive. In this paper, we propose a one-stage network for real-time PNG, termed End-to-End Panoptic Narrative Grounding network (EPNG), which directly generates masks for referents. Specifically, we propose two innovative designs, i.e., Locality-Perceptive Attention (LPA) and a bidirectional Semantic Alignment Loss (SAL), to properly handle the many-to-many relationship between textual expressions and visual objects. LPA embeds the local spatial priors into attention modeling, i.e., a pixel may belong to multiple masks at different scales, thereby improving segmentation. To help understand the complex semantic relationships, SAL proposes a bidirectional contrastive objective to regularize the semantic consistency inter modalities. Extensive experiments on the PNG benchmark dataset demonstrate the effectiveness and efficiency of our method. Compared to the single-stage baseline, our method achieves a significant improvement of up to 9.4% accuracy. More importantly, our EPNG is 10 times faster than the two-stage model. Meanwhile, the generalization ability of EPNG is also validated by zero-shot experiments on other grounding tasks.

Jing He, Yiyi Zhou, Qi Zhang et al.

Pixel synthesis is a promising research paradigm for image generation, which can well exploit pixel-wise prior knowledge for generation. However, existing methods still suffer from excessive memory footprint and computation overhead. In this paper, we propose a progressive pixel synthesis network towards efficient image generation, coined as PixelFolder. Specifically, PixelFolder formulates image generation as a progressive pixel regression problem and synthesizes images via a multi-stage structure, which can greatly reduce the overhead caused by large tensor transformations. In addition, we introduce novel pixel folding operations to further improve model efficiency while maintaining pixel-wise prior knowledge for end-to-end regression. With these innovative designs, we greatly reduce the expenditure of pixel synthesis, e.g., reducing 89% computation and 53% parameters compared with the latest pixel synthesis method CIPS. To validate our approach, we conduct extensive experiments on two benchmark datasets, namely FFHQ and LSUN Church. The experimental results show that with much less expenditure, PixelFolder obtains new state-of-the-art (SOTA) performance on two benchmark datasets, i.e., 3.77 FID and 2.45 FID on FFHQ and LSUN Church, respectively.Meanwhile, PixelFolder is also more efficient than the SOTA methods like StyleGAN2, reducing about 72% computation and 31% parameters, respectively. These results greatly validate the effectiveness of the proposed PixelFolder.

Qiong Wu, Shubin Huang, Yiyi Zhou et al.

Prompt tuning is a parameter-efficient way to deploy large-scale pre-trained models to downstream tasks by adding task-specific tokens. In terms of vision-language pre-trained (VLP) models, prompt tuning often requires a large number of learnable tokens to bridge the gap between the pre-training and downstream tasks, which greatly exacerbates the already high computational overhead. In this paper, we revisit the principle of prompt tuning for Transformer-based VLP models, and reveal that the impact of soft prompt tokens can be actually approximated via independent information diffusion steps, thereby avoiding the expensive global attention modeling and reducing the computational complexity to a large extent. Based on this finding, we propose a novel Approximated Prompt Tuning (APT) approach towards efficient VL transfer learning. To validate APT, we apply it to two representative VLP models, namely ViLT and METER, and conduct extensive experiments on a bunch of downstream tasks. Meanwhile, the generalization of APT is also validated on CLIP for image classification and StableDiffusion for text-to-image generation. The experimental results not only show the superior performance gains and computation efficiency of APT against the conventional prompt tuning methods, e.g., +7.01% accuracy and -82.30% additional computation overhead on METER, but also confirm its merits over other parameter-efficient transfer learning approaches.

Shubin Huang, Qiong Wu, Yiyi Zhou et al.

Pre-trained language models (PLMs) have played an increasing role in multimedia research. In terms of vision-language (VL) tasks, they often serve as a language encoder and still require an additional fusion network for VL reasoning, resulting in excessive memory overhead. In this paper, we focus on exploring PLMs as a stand-alone model for VL reasoning tasks. Inspired by the recently popular prompt tuning, we first prove that the processed visual features can be also projected onto the semantic space of PLMs and act as prompt tokens to bridge the gap between single- and multi-modal learning. However, this solution exhibits obvious redundancy in visual information and model inference, and the placement of prompt tokens also greatly affects the final performance. Based on these observations, we further propose a novel transfer learning approach for PLMs, termed Dynamic Visual Prompting (DVP). Concretely, DVP first deploys a cross-attention module to obtain text-related and compact visual prompt tokens, thereby greatly reducing the input length of PLMs. To obtain the optimal placement, we also equip DVP with a reinforcement-learning based search algorithm, which can automatically merge DVP with PLMs for different VL tasks via a very short search process. In addition, we also experiment DVP with the recently popular adapter approach to keep the most parameters of PLMs intact when adapting to VL tasks, helping PLMs achieve a quick shift between single- and multi-modal tasks. We apply DVP to two representative PLMs, namely BERT and T5, and conduct extensive experiments on a set of VL reasoning benchmarks including VQA2.0, GQA and SNLIVE. The experimental results not only show the advantage of DVP on efficiency and performance, but also confirm its superiority in adapting pre-trained language models to VL tasks.

Qiong Wu, Jiaer Xia, Pingyang Dai et al.

Visible-infrared person re-identification (VI-ReID) is a task of matching the same individuals across the visible and infrared modalities. Its main challenge lies in the modality gap caused by cameras operating on different spectra. Existing VI-ReID methods mainly focus on learning general features across modalities, often at the expense of feature discriminability. To address this issue, we present a novel cycle-construction-based network for neutral yet discriminative feature learning, termed CycleTrans. Specifically, CycleTrans uses a lightweight Knowledge Capturing Module (KCM) to capture rich semantics from the modality-relevant feature maps according to pseudo queries. Afterwards, a Discrepancy Modeling Module (DMM) is deployed to transform these features into neutral ones according to the modality-irrelevant prototypes. To ensure feature discriminability, another two KCMs are further deployed for feature cycle constructions. With cycle construction, our method can learn effective neutral features for visible and infrared images while preserving their salient semantics. Extensive experiments on SYSU-MM01 and RegDB datasets validate the merits of CycleTrans against a flurry of state-of-the-art methods, +4.57% on rank-1 in SYSU-MM01 and +2.2% on rank-1 in RegDB.

Tao Chen, Kun Zhang, Qiong Wu et al.

Long video understanding is a key challenge that plagues the advancement of \emph{Multimodal Large language Models} (MLLMs). In this paper, we study this problem from the perspective of visual memory mechanism, and proposed a novel and training-free approach, termed \emph{Flexible Memory} (\textbf{FlexMem}). In principle, FlexMem aims to mimic human behavior of video watching, \emph{i.e.}, continually watching video content and recalling the most relevant memory fragments to answer the question. In this way, FlexMem can help MLLMs achieve video understanding of infinite lengths, unlike previous methods that process all video information at once and have input upper-limit. Concretely, FlexMem first consider the visual KV caches as the memory sources, and realize the effective memory transfer and writing via a dual-pathway compression design. Afterwards, FlexMem also explores different memory reading strategies for the diverse video understanding tasks, including the popular streaming one. To validate FlexMem, we apply it to two popular video-MLLMs, and conduct extensive experiments on five long video and one streaming video task. The experimental results show that on \textbf{a single 3090 GPU}, our FlexMem can achieve obvious improvements than existing efficient video understanding methods and process more than \textbf{1k frames}, which also helps the base MLLMs achieve comparable or even better performance than SOTA MLLMs on some benchmarks, \emph{e.g.} , GPT-4o and Gemini-1.5 Pro.

Minglang Huang, Yiyi Zhou, Gen Luo et al.

Referring Expression Segmentation (RES) is an emerging task in computer vision, which segments the target instances in images based on text descriptions. However, its development is plagued by the expensive segmentation labels. To address this issue, we propose a new learning task for RES called Omni-supervised Referring Expression Segmentation (Omni-RES), which aims to make full use of unlabeled, fully labeled and weakly labeled data, e.g., referring points or grounding boxes, for efficient RES training. To accomplish this task, we also propose a novel yet strong baseline method for Omni-RES based on the recently popular teacher-student learning, where the weak labels are not directly transformed into supervision signals but used as a yardstick to select and refine high-quality pseudo-masks for teacher-student learning. To validate the proposed Omni-RES method, we apply it to a set of state-of-the-art RES models and conduct extensive experiments on a bunch of RES datasets. The experimental results yield the obvious merits of Omni-RES than the fully-supervised and semi-supervised training schemes. For instance, with only 10% fully labeled data, Omni-RES can help the base model achieve 100% fully supervised performance, and it also outperform the semi-supervised alternative by a large margin, e.g., +14.93% on RefCOCO and +14.95% on RefCOCO+, respectively. More importantly, Omni-RES also enable the use of large-scale vision-langauges like Visual Genome to facilitate low-cost RES training, and achieve new SOTA performance of RES, e.g., 80.66 on RefCOCO.

Gen Luo, Yiyi Zhou, Yuxin Zhang et al.

Despite remarkable progress, existing multimodal large language models (MLLMs) are still inferior in granular visual recognition. Contrary to previous works, we study this problem from the perspective of image resolution, and reveal that a combination of low- and high-resolution visual features can effectively mitigate this shortcoming. Based on this observation, we propose a novel and efficient method for MLLMs, termed Mixture-of-Resolution Adaptation (MRA). In particular, MRA adopts two visual pathways for images with different resolutions, where high-resolution visual information is embedded into the low-resolution pathway via the novel mixture-of-resolution adapters (MR-Adapters). This design also greatly reduces the input sequence length of MLLMs. To validate MRA, we apply it to a recent MLLM called LLaVA, and term the new model LLaVA-HR. We conduct extensive experiments on 11 vision-language (VL) tasks, which show that LLaVA-HR outperforms existing MLLMs on 8 VL tasks, e.g., +9.4% on TextVQA. More importantly, both training and inference of LLaVA-HR remain efficient with MRA, e.g., 20 training hours and 3$\times$ inference speed than LLaVA-1.5. Source codes are released at: https://github.com/luogen1996/LLaVA-HR.

Tao Chen, Shaobo Ju, Qiong Wu et al.

Due to excessive memory overhead, most Multimodal Large Language Models (MLLMs) can only process videos of limited frames. In this paper, we propose an effective and efficient paradigm to remedy this shortcoming, termed One-shot video-Clip based Retrieval AuGmentation (OneClip-RAG). Compared with existing video RAG methods, OneClip-RAG makes full use of the merits of video clips for augmented video understanding in terms of both knowledge integrity and semantic coherence. Besides, it is also equipped with a novel query-guided video chunking algorithm that can unify clip chunking and cross-modal retrieval in one processing step, avoiding redundant computations. To improve instruction following, we further propose a new dataset called SynLongVideo and design a progressive training regime for OneClip-RAG. OneClip-RAG is plugged into five recent MLLMs and validated on a set of long-video benchmarks. Experimental results not only show the obvious performance gains by OneClip-RAG over MLLMs, e.g., boosting InternLV2 8B and Qwen2-VL 7B to the level of GPT-4o on MLVU, but also show its superior efficiency in handling long videos. e.g., enabling LLaVA-Video understand up to an hour of videos in less than 2.2 minutes on a single 4090 GPU.

Qiancheng Zheng, Yunhang Shen, Gen Luo et al.

In this paper, we propose a novel task termed Omni-Referring Image Segmentation (OmniRIS) towards highly generalized image segmentation. Compared with existing unimodally conditioned segmentation tasks, such as RIS and visual RIS, OmniRIS supports the input of text instructions and reference images with masks, boxes or scribbles as omni-prompts. This property makes it can well exploit the intrinsic merits of both text and visual modalities, i.e., granular attribute referring and uncommon object grounding, respectively. Besides, OmniRIS can also handle various segmentation settings, such as one v.s. many and many v.s. many, further facilitating its practical use. To promote the research of OmniRIS, we also rigorously design and construct a large dataset termed OmniRef, which consists of 186,939 omni-prompts for 30,956 images, and establish a comprehensive evaluation system. Moreover, a strong and general baseline termed OmniSegNet is also proposed to tackle the key challenges of OmniRIS, such as omni-prompt encoding. The extensive experiments not only validate the capability of OmniSegNet in following omni-modal instructions, but also show the superiority of OmniRIS for highly generalized image segmentation.

Yunhang Shen, Chaoyou Fu, Shaoqi Dong et al.

We introduce Long-VITA, a simple yet effective large multi-modal model for long-context visual-language understanding tasks. It is adept at concurrently processing and analyzing modalities of image, video, and text over 4K frames or 1M tokens while delivering advanced performances on short-context multi-modal tasks. We propose an effective multi-modal training schema that starts with large language models and proceeds through vision-language alignment, general knowledge learning, and two sequential stages of long-sequence fine-tuning. We further implement context-parallelism distributed inference and logits-masked language modeling head to scale Long-VITA to infinitely long inputs of images and texts during model inference. Regarding training data, Long-VITA is built on a mix of 17M samples from public datasets only and demonstrates state-of-the-art performance on various multi-modal benchmarks, compared against recent cutting-edge models with internal data. Long-VITA is fully open-source and reproducible.. By leveraging our inference designs, Long-VITA models achieve a remarkable 2x prefill speedup and 4x context length extension in a single node with 8 GPUs. We hope Long-VITA can serve as a competitive baseline and offer valuable insights for the open-source community in advancing long-context multi-modal understanding.

Jinlu Zhang, Yiyi Zhou, Qiancheng Zheng et al.

Text-to-3D-aware face (T3D Face) generation and manipulation is an emerging research hot spot in machine learning, which still suffers from low efficiency and poor quality. In this paper, we propose an End-to-End Efficient and Effective network for fast and accurate T3D face generation and manipulation, termed $E^3$-FaceNet. Different from existing complex generation paradigms, $E^3$-FaceNet resorts to a direct mapping from text instructions to 3D-aware visual space. We introduce a novel Style Code Enhancer to enhance cross-modal semantic alignment, alongside an innovative Geometric Regularization objective to maintain consistency across multi-view generations. Extensive experiments on three benchmark datasets demonstrate that $E^3$-FaceNet can not only achieve picture-like 3D face generation and manipulation, but also improve inference speed by orders of magnitudes. For instance, compared with Latent3D, $E^3$-FaceNet speeds up the five-view generations by almost 470 times, while still exceeding in generation quality. Our code is released at https://github.com/Aria-Zhangjl/E3-FaceNet.

Qiong Wu, Wenhao Lin, Yiyi Zhou et al.

The excessive use of visual tokens in existing Multimoal Large Language Models (MLLMs) often exhibits obvious redundancy and brings in prohibitively expensive computation. To gain insights into this problem, we first conduct extensive empirical studies on the attention behaviors of MLLMs, and summarize three main inference stages in MLLMs: (i) Early fusion between tokens is first accomplished quickly. (ii) Intra-modality modeling then comes to play. (iii) Multimodal reasoning} resumes and lasts until the end of inference. In particular, we reveal that visual tokens will stop contributing to reasoning when the text tokens receive enough image information, yielding obvious visual redundancy. Based on these generalized observations, we propose a simple yet effective method to improve the efficiency of MLLMs, termed dynamic visual-token exit (DyVTE). DyVTE uses lightweight hyper-networks to perceive the text token status and decide the removal of all visual tokens after a certain layer, thereby addressing the observed visual redundancy. To validate VTE, we apply it to a set of MLLMs, including LLaVA, VILA, Eagle and InternVL, and conduct extensive experiments on a bunch of benchmarks. The experiment results not only show the effectiveness of our VTE in improving MLLMs' efficiency, but also yield the general modeling patterns of MLLMs, well facilitating the in-depth understanding of MLLMs. Our code is released at https://github.com/DoubtedSteam/DyVTE.

Shuheng Zhang, Yuqi Liu, Hongbo Zhou et al.

Despite great progress, text-driven long video editing is still notoriously challenging mainly due to excessive memory overhead. Although recent efforts have simplified this task into a two-step process of keyframe translation and interpolation generation, the token-wise keyframe translation still plagues the upper limit of video length. In this paper, we propose a novel and training-free approach towards efficient and effective long video editing, termed AdaFlow. We first reveal that not all tokens of video frames hold equal importance for keyframe translation, based on which we propose an Adaptive Attention Slimming scheme for AdaFlow to squeeze the $KV$ sequence, thus increasing the number of keyframes for translations by an order of magnitude. In addition, an Adaptive Keyframe Selection scheme is also equipped to select the representative frames for joint editing, further improving generation quality. With these innovative designs, AdaFlow achieves high-quality long video editing of minutes in one inference, i.e., more than 1$k$ frames on one A800 GPU, which is about ten times longer than the compared methods, e.g., TokenFlow. To validate AdaFlow, we also build a new benchmark for long video editing with high-quality annotations, termed LongV-EVAL. Our code is released at: https://github.com/jidantang55/AdaFlow.

Xiaorui Huang, Gen Luo, Chaoyang Zhu et al.

Recently, Segment Anything Model (SAM) has become a research hotspot in the fields of multimedia and computer vision, which exhibits powerful yet versatile capabilities on various (un) conditional image segmentation tasks. Although SAM can support different types of segmentation prompts, we note that, compared to point- and box-guided segmentations, it performs much worse on text-instructed tasks, e.g., referring image segmentation (RIS). In this paper, we argue that deep text instruction tuning is key to mitigate such shortcoming caused by the shallow fusion scheme in its default light-weight mask decoder. To address this issue, we propose two simple yet effective deep instruction tuning (DIT) methods for SAM, one is end-to-end and the other is layer-wise. With minimal modifications, DITs can directly transform the image encoder of SAM as a stand-alone vision-language learner in contrast to building another deep fusion branch, maximizing the benefit of its superior segmentation capability. Extensive experiments on three highly competitive benchmark datasets of RIS show that a simple end-to-end DIT can improve SAM by a large margin, while the layer-wise DIT can further boost the performance to state-of-the-art with much less data and training expenditures. Our code is released at: https://github.com/wysnzzzz/DIT.

Ziyue Zhang, Luxi Lin, Xiaolin Hu et al.

Diffusion inversion is a task of recovering the noise of an image in a diffusion model, which is vital for controllable diffusion image editing. At present, diffusion inversion still remains a challenging task due to the lack of viable supervision signals. Thus, most existing methods resort to approximation-based solutions, which however are often at the cost of performance or efficiency. To remedy these shortcomings, we propose a novel self-supervised diffusion inversion approach in this paper, termed Deep Inversion (DeepInv). Instead of requiring ground-truth noise annotations, we introduce a self-supervised objective as well as a data augmentation strategy to generate high-quality pseudo noises from real images without manual intervention. Based on these two innovative designs, DeepInv is also equipped with an iterative and multi-scale training regime to train a parameterized inversion solver, thereby achieving the fast and accurate image-to-noise mapping. To the best of our knowledge, this is the first attempt of presenting a trainable solver to predict inversion noise step by step. The extensive experiments show that our DeepInv can achieve much better performance and inference speed than the compared methods, e.g., +40.435% SSIM than EasyInv and +9887.5% speed than ReNoise on COCO dataset. Moreover, our careful designs of trainable solvers can also provide insights to the community. Codes and model parameters will be released in https://github.com/potato-kitty/DeepInv.

Zhiyao Wang, Xu Chen, Chengming Xu et al.

Face Restoration (FR) is a crucial area within image and video processing, focusing on reconstructing high-quality portraits from degraded inputs. Despite advancements in image FR, video FR remains relatively under-explored, primarily due to challenges related to temporal consistency, motion artifacts, and the limited availability of high-quality video data. Moreover, traditional face restoration typically prioritizes enhancing resolution and may not give as much consideration to related tasks such as facial colorization and inpainting. In this paper, we propose a novel approach for the Generalized Video Face Restoration (GVFR) task, which integrates video BFR, inpainting, and colorization tasks that we empirically show to benefit each other. We present a unified framework, termed as stable video face restoration (SVFR), which leverages the generative and motion priors of Stable Video Diffusion (SVD) and incorporates task-specific information through a unified face restoration framework. A learnable task embedding is introduced to enhance task identification. Meanwhile, a novel Unified Latent Regularization (ULR) is employed to encourage the shared feature representation learning among different subtasks. To further enhance the restoration quality and temporal stability, we introduce the facial prior learning and the self-referred refinement as auxiliary strategies used for both training and inference. The proposed framework effectively combines the complementary strengths of these tasks, enhancing temporal coherence and achieving superior restoration quality. This work advances the state-of-the-art in video FR and establishes a new paradigm for generalized video face restoration. Code and video demo are available at https://github.com/wangzhiyaoo/SVFR.git.

Chaoyang Zhu, Yiyi Zhou, Yunhang Shen et al.

In this paper, we propose a simple yet universal network termed SeqTR for visual grounding tasks, e.g., phrase localization, referring expression comprehension (REC) and segmentation (RES). The canonical paradigms for visual grounding often require substantial expertise in designing network architectures and loss functions, making them hard to generalize across tasks. To simplify and unify the modeling, we cast visual grounding as a point prediction problem conditioned on image and text inputs, where either the bounding box or binary mask is represented as a sequence of discrete coordinate tokens. Under this paradigm, visual grounding tasks are unified in our SeqTR network without task-specific branches or heads, e.g., the convolutional mask decoder for RES, which greatly reduces the complexity of multi-task modeling. In addition, SeqTR also shares the same optimization objective for all tasks with a simple cross-entropy loss, further reducing the complexity of deploying hand-crafted loss functions. Experiments on five benchmark datasets demonstrate that the proposed SeqTR outperforms (or is on par with) the existing state-of-the-arts, proving that a simple yet universal approach for visual grounding is indeed feasible. Source code is available at https://github.com/sean-zhuh/SeqTR.

Qiong Wu, Xiangcong Yang, Yiyi Zhou et al.

Despite great progress, existing multimodal large language models (MLLMs) are prone to visual hallucination, greatly impeding their trustworthy applications. In this paper, we study this problem from the perspective of visual-spatial reasoning, and propose a new learning task for MLLMs, termed Grounded Chain-of-Thought (GCoT). Different from recent visual CoT studies, which focus more on visual knowledge reasoning, GCoT is keen to helping MLLMs to recognize and ground the relevant visual cues step by step, thereby predicting the correct answer with grounding coordinates as the intuitive basis. To facilitate this task, we also carefully design and construct a dataset called multimodal grounded chain-of-thought (MM-GCoT) consisting of 24,022 GCoT examples for 5,033 images. Besides, a comprehensive consistency evaluation system is also introduced, including the metrics of answer accuracy, grounding accuracy and answer-grounding consistency. We further design and conduct a bunch of experiments on 12 advanced MLLMs, and reveal some notable findings: i. most MLLMs performs poorly on the consistency evaluation, indicating obvious visual hallucination; ii. visual hallucination is not directly related to the parameter size and general multimodal performance, i.e., a larger and stronger MLLM is not less affected by this issue. Lastly, we also demonstrate that the proposed dataset can help existing MLLMs to well cultivate their GCoT capability and reduce the inconsistent answering significantly. Moreover, their GCoT can be also generalized to exiting multimodal tasks, such as open-world QA and REC.

Yutao Jiang, Qiong Wu, Wenhao Lin et al.

Recent Multimodal Large Language Models(MLLMs) often use a large number of visual tokens to compensate their visual shortcoming, leading to excessive computation and obvious visual redundancy. In this paper, we investigate what kind of visual tokens are needed for MLLMs, and reveal that both foreground and background tokens are critical for MLLMs given the varying difficulties of examples. Based on this observation, we propose a graph-based method towards training-free visual token pruning, termed G-Prune.In particular, G-Prune regards visual tokens as nodes, and construct their connections based on their semantic similarities. Afterwards, the information flow is propagated via weighted links, and the most important tokens after iterations are kept for MLLMs, which can be front or background.To validate G-Prune, we apply it to a recent MLLM called LLaVA-NeXT, and conduct extensive experiments on a set of benchmarks.The experiment results show that G-Prune can greatly reduce computation overhead while retaining high performance on both coarse- and fine-grained tasks. For instance, G-Prune can reduce 63.57\% FLOPs of LLaVA-NeXT on VQA2.0 and TextVQA with only 0.95\% and 2.34\% accuracy drops, respectively.

Siyu Zou, Jiji Tang, Yiyi Zhou et al.

Diffusion-based Image Editing (DIE) is an emerging research hot-spot, which often applies a semantic mask to control the target area for diffusion-based editing. However, most existing solutions obtain these masks via manual operations or off-line processing, greatly reducing their efficiency. In this paper, we propose a novel and efficient image editing method for Text-to-Image (T2I) diffusion models, termed Instant Diffusion Editing(InstDiffEdit). In particular, InstDiffEdit aims to employ the cross-modal attention ability of existing diffusion models to achieve instant mask guidance during the diffusion steps. To reduce the noise of attention maps and realize the full automatics, we equip InstDiffEdit with a training-free refinement scheme to adaptively aggregate the attention distributions for the automatic yet accurate mask generation. Meanwhile, to supplement the existing evaluations of DIE, we propose a new benchmark called Editing-Mask to examine the mask accuracy and local editing ability of existing methods. To validate InstDiffEdit, we also conduct extensive experiments on ImageNet and Imagen, and compare it with a bunch of the SOTA methods. The experimental results show that InstDiffEdit not only outperforms the SOTA methods in both image quality and editing results, but also has a much faster inference speed, i.e., +5 to +6 times.

Yaxin Luo, Gen Luo, Jiayi Ji et al.

Despite the significant progress in multimodal large language models (MLLMs), their high computational cost remains a barrier to real-world deployment. Inspired by the mixture of depths (MoDs) in natural language processing, we aim to address this limitation from the perspective of ``activated tokens''. Our key insight is that if most tokens are redundant for the layer computation, then can be skipped directly via the MoD layer. However, directly converting the dense layers of MLLMs to MoD layers leads to substantial performance degradation. To address this issue, we propose an innovative MoD adaptation strategy for existing MLLMs called $γ$-MoD. In $γ$-MoD, a novel metric is proposed to guide the deployment of MoDs in the MLLM, namely rank of attention maps (ARank). Through ARank, we can effectively identify which layer is redundant and should be replaced with the MoD layer. Based on ARank, we further propose two novel designs to maximize the computational sparsity of MLLM while maintaining its performance, namely shared vision-language router and masked routing learning. With these designs, more than 90% dense layers of the MLLM can be effectively converted to the MoD ones. To validate our method, we apply it to three popular MLLMs, and conduct extensive experiments on 9 benchmark datasets. Experimental results not only validate the significant efficiency benefit of $γ$-MoD to existing MLLMs but also confirm its generalization ability on various MLLMs. For example, with a minor performance drop, i.e., -1.5%, $γ$-MoD can reduce the training and inference time of LLaVA-HR by 31.0% and 53.2%, respectively.

Bo Tong, Bokai Lai, Yiyi Zhou et al.

Despite a big leap forward in capability, multimodal large language models (MLLMs) tend to behave like a sloth in practical use, i.e., slow response and large latency. Recent efforts are devoted to building tiny MLLMs for better efficiency, but the plethora of visual tokens still used limit their actual speedup. In this paper, we propose a powerful and fast tiny MLLM called FlashSloth. Different from previous efforts, FlashSloth focuses on improving the descriptive power of visual tokens in the process of compressing their redundant semantics. In particular, FlashSloth introduces embedded visual compression designs to capture both visually salient and instruction-related image information, so as to achieving superior multimodal performance with fewer visual tokens. Extensive experiments are conducted to validate the proposed FlashSloth, and a bunch of tiny but strong MLLMs are also comprehensively compared, e.g., InternVL2, MiniCPM-V2 and Qwen2-VL. The experimental results show that compared with these advanced tiny MLLMs, our FlashSloth can greatly reduce the number of visual tokens, training memory and computation complexity while retaining high performance on various VL tasks.

Shaobo Ju, Baiyang Song, Tao Chen et al.

Due to the great saving of computation and memory overhead, token compression has become a research hot-spot for MLLMs and achieved remarkable progress in image-language tasks. However, for the video, existing methods still fall short of high-ratio token compression. We attribute this shortcoming to the insufficient modeling of temporal and continual video content, and propose a novel and training-free token pruning method for video MLLMs, termed ForestPrune, which achieves effective and high-ratio pruning via Spatial-temporal Forest Modeling. In practice, ForestPrune construct token forests across video frames based on the semantic, spatial and temporal constraints, making an overall comprehension of videos. Afterwards, ForestPrune evaluates the importance of token trees and nodes based on tree depth and node roles, thereby obtaining a globally optimal pruning decision. To validate ForestPrune, we apply it to two representative video MLLMs, namely LLaVA-Video and LLaVA-OneVision, and conduct extensive experiments on a bunch of video benchmarks. The experimental results not only show the great effectiveness for video MLLMs, e.g., retaining 95.8% average accuracy while reducing 90% tokens for LLaVA-OneVision, but also show its superior performance and efficiency than the compared token compression methods, e.g., +10.1% accuracy on MLVU and -81.4% pruning time than FrameFusion on LLaVA-Video.

Qiong Wu, Weihao Ye, Yiyi Zhou et al.

In this paper, we propose a novel parameter and computation efficient tuning method for Multi-modal Large Language Models (MLLMs), termed Efficient Attention Skipping (EAS). Concretely, we first reveal that multi-head attentions (MHAs), the main computational overhead of MLLMs, are often redundant to downstream tasks. Based on this observation, EAS evaluates the attention redundancy and skips the less important MHAs to speed up inference. Besides, we also propose a novel propagation-of-information adapter (PIA) to serve the attention skipping of EAS and keep parameter efficiency, which can be further re-parameterized into feed-forward networks (FFNs) for zero-extra latency. To validate EAS, we apply it to a recently proposed MLLM called LaVIN and a classic VL pre-trained model called METER, and conduct extensive experiments on a set of benchmarks. The experiments show that EAS not only retains high performance and parameter efficiency, but also greatly speeds up inference speed. For instance, LaVIN-EAS can obtain 89.98\% accuracy on ScineceQA while speeding up inference by 2.2 times to LaVIN

Yiwei Ma, Jiayi Ji, Xiaoshuai Sun et al.

This paper explores a novel dynamic network for vision and language tasks, where the inferring structure is customized on the fly for different inputs. Most previous state-of-the-art approaches are static and hand-crafted networks, which not only heavily rely on expert knowledge, but also ignore the semantic diversity of input samples, therefore resulting in suboptimal performance. To address these issues, we propose a novel Dynamic Transformer Network (DTNet) for image captioning, which dynamically assigns customized paths to different samples, leading to discriminative yet accurate captions. Specifically, to build a rich routing space and improve routing efficiency, we introduce five types of basic cells and group them into two separate routing spaces according to their operating domains, i.e., spatial and channel. Then, we design a Spatial-Channel Joint Router (SCJR), which endows the model with the capability of path customization based on both spatial and channel information of the input sample. To validate the effectiveness of our proposed DTNet, we conduct extensive experiments on the MS-COCO dataset and achieve new state-of-the-art performance on both the Karpathy split and the online test server.

Qiong Wu, Wei Yu, Yiyi Zhou et al.

With ever increasing parameters and computation, vision-language pre-trained (VLP) models exhibit prohibitive expenditure in downstream task adaption. Recent endeavors mainly focus on parameter efficient transfer learning (PETL) for VLP models by only updating a small number of parameters. However, excessive computational overhead still plagues the application of VLPs. In this paper, we aim at parameter and computation efficient transfer learning (PCETL) for VLP models. In particular, PCETL not only needs to limit the number of trainable parameters in VLP models, but also to reduce the computational redundancy during inference, thus enabling a more efficient transfer. To approach this target, we propose a novel dynamic architecture skipping (DAS) approach towards effective PCETL. Instead of directly optimizing the intrinsic architectures of VLP models, DAS first observes the significances of their modules to downstream tasks via a reinforcement learning (RL) based process, and then skips the redundant ones with lightweight networks, i.e., adapters, according to the obtained rewards. In this case, the VLP model can well maintain the scale of trainable parameters while speeding up its inference on downstream tasks. To validate DAS, we apply it to two representative VLP models, namely ViLT and METER, and conduct extensive experiments on a bunch of VL tasks. The experimental results not only show the great advantages of DAS in reducing computational complexity, e.g. -11.97% FLOPs of METER on VQA2.0, but also confirm its competitiveness against existing PETL methods in terms of parameter scale and performance. Our source code is given in our appendix.

Gen Luo, Yiyi Zhou, Tianhe Ren et al.

Recently, growing interest has been aroused in extending the multimodal capability of large language models (LLMs), e.g., vision-language (VL) learning, which is regarded as the next milestone of artificial general intelligence. However, existing solutions are prohibitively expensive, which not only need to optimize excessive parameters, but also require another large-scale pre-training before VL instruction tuning. In this paper, we propose a novel and affordable solution for the effective VL adaption of LLMs, called Mixture-of-Modality Adaptation (MMA). Instead of using large neural networks to connect the image encoder and LLM, MMA adopts lightweight modules, i.e., adapters, to bridge the gap between LLMs and VL tasks, which also enables the joint optimization of the image and language models. Meanwhile, MMA is also equipped with a routing algorithm to help LLMs achieve an automatic shift between single- and multi-modal instructions without compromising their ability of natural language understanding. To validate MMA, we apply it to a recent LLM called LLaMA and term this formed large vision-language instructed model as LaVIN. To validate MMA and LaVIN, we conduct extensive experiments under two setups, namely multimodal science question answering and multimodal dialogue. The experimental results not only demonstrate the competitive performance and the superior training efficiency of LaVIN than existing multimodal LLMs, but also confirm its great potential as a general-purpose chatbot. More importantly, the actual expenditure of LaVIN is extremely cheap, e.g., only 1.4 training hours with 3.8M trainable parameters, greatly confirming the effectiveness of MMA. Our project is released at https://luogen1996.github.io/lavin.

Gen Luo, Yiyi Zhou, Xiaoshuai Sun et al.

In this paper, we are committed to establishing an unified and end-to-end multi-modal network via exploring the language-guided visual recognition. To approach this target, we first propose a novel multi-modal convolution module called Language-dependent Convolution (LaConv). Its convolution kernels are dynamically generated based on natural language information, which can help extract differentiated visual features for different multi-modal examples. Based on the LaConv module, we further build the first fully language-driven convolution network, termed as LaConvNet, which can unify the visual recognition and multi-modal reasoning in one forward structure. To validate LaConv and LaConvNet, we conduct extensive experiments on four benchmark datasets of two vision-and-language tasks, i.e., visual question answering (VQA) and referring expression comprehension (REC). The experimental results not only shows the performance gains of LaConv compared to the existing multi-modal modules, but also witness the merits of LaConvNet as an unified network, including compact network, high generalization ability and excellent performance, e.g., +4.7% on RefCOCO+.

Gen Luo, Yiyi Zhou, Xiaoshuai Sun et al.

Referring expression comprehension (REC) and segmentation (RES) are two highly-related tasks, which both aim at identifying the referent according to a natural language expression. In this paper, we propose a novel Multi-task Collaborative Network (MCN) to achieve a joint learning of REC and RES for the first time. In MCN, RES can help REC to achieve better language-vision alignment, while REC can help RES to better locate the referent. In addition, we address a key challenge in this multi-task setup, i.e., the prediction conflict, with two innovative designs namely, Consistency Energy Maximization (CEM) and Adaptive Soft Non-Located Suppression (ASNLS). Specifically, CEM enables REC and RES to focus on similar visual regions by maximizing the consistency energy between two tasks. ASNLS supresses the response of unrelated regions in RES based on the prediction of REC. To validate our model, we conduct extensive experiments on three benchmark datasets of REC and RES, i.e., RefCOCO, RefCOCO+ and RefCOCOg. The experimental results report the significant performance gains of MCN over all existing methods, i.e., up to +7.13% for REC and +11.50% for RES over SOTA, which well confirm the validity of our model for joint REC and RES learning.

Yiyi Zhou, Rongrong Ji, Gen Luo et al.

Referring Expression Comprehension (REC) is an emerging research spot in computer vision, which refers to detecting the target region in an image given an text description. Most existing REC methods follow a multi-stage pipeline, which are computationally expensive and greatly limit the application of REC. In this paper, we propose a one-stage model towards real-time REC, termed Real-time Global Inference Network (RealGIN). RealGIN addresses the diversity and complexity issues in REC with two innovative designs: the Adaptive Feature Selection (AFS) and the Global Attentive ReAsoNing unit (GARAN). AFS adaptively fuses features at different semantic levels to handle the varying content of expressions. GARAN uses the textual feature as a pivot to collect expression-related visual information from all regions, and thenselectively diffuse such information back to all regions, which provides sufficient context for modeling the complex linguistic conditions in expressions. On five benchmark datasets, i.e., RefCOCO, RefCOCO+, RefCOCOg, ReferIt and Flickr30k, the proposed RealGIN outperforms most prior works and achieves very competitive performances against the most advanced method, i.e., MAttNet. Most importantly, under the same hardware, RealGIN can boost the processing speed by about 10 times over the existing methods.