Bin Ren

Xingyu Qiu, Yuqian Fu, Jiawei Geng et al.

Cross-domain few-shot object detection (CD-FSOD) remains a challenging problem for existing object detectors and few-shot learning approaches, particularly when generalizing across distinct domains. As part of NTIRE 2026, we hosted the second CD-FSOD Challenge to systematically evaluate and promote progress in detecting objects in unseen target domains under limited annotation conditions. The challenge received strong community interest, with 128 registered participants and a total of 696 submissions. Among them, 31 teams actively participated, and 19 teams submitted valid final results. Participants explored a wide range of strategies, introducing innovative methods that push the performance frontier under both open-source and closed-source tracks. This report presents a detailed overview of the NTIRE 2026 CD-FSOD Challenge, including a summary of the submitted approaches and an analysis of the final results across all participating teams. Challenge Codes: https://github.com/ohMargin/NTIRE2026_CDFSOD.

Bin Ren, Yahui Liu, Yue Song et al.

Position Embeddings (PEs), an arguably indispensable component in Vision Transformers (ViTs), have been shown to improve the performance of ViTs on many vision tasks. However, PEs have a potentially high risk of privacy leakage since the spatial information of the input patches is exposed. This caveat naturally raises a series of interesting questions about the impact of PEs on the accuracy, privacy, prediction consistency, etc. To tackle these issues, we propose a Masked Jigsaw Puzzle (MJP) position embedding method. In particular, MJP first shuffles the selected patches via our block-wise random jigsaw puzzle shuffle algorithm, and their corresponding PEs are occluded. Meanwhile, for the non-occluded patches, the PEs remain the original ones but their spatial relation is strengthened via our dense absolute localization regressor. The experimental results reveal that 1) PEs explicitly encode the 2D spatial relationship and lead to severe privacy leakage problems under gradient inversion attack; 2) Training ViTs with the naively shuffled patches can alleviate the problem, but it harms the accuracy; 3) Under a certain shuffle ratio, the proposed MJP not only boosts the performance and robustness on large-scale datasets (i.e., ImageNet-1K and ImageNet-C, -A/O) but also improves the privacy preservation ability under typical gradient attacks by a large margin. The source code and trained models are available at~\url{https://github.com/yhlleo/MJP}.

Bin Ren, Hao Tang, Yiming Wang et al.

For semantic-guided cross-view image translation, it is crucial to learn where to sample pixels from the source view image and where to reallocate them guided by the target view semantic map, especially when there is little overlap or drastic view difference between the source and target images. Hence, one not only needs to encode the long-range dependencies among pixels in both the source view image and target view semantic map but also needs to translate these learned dependencies. To this end, we propose a novel generative adversarial network, PI-Trans, which mainly consists of a novel Parallel-ConvMLP module and an Implicit Transformation module at multiple semantic levels. Extensive experimental results show that PI-Trans achieves the best qualitative and quantitative performance by a large margin compared to the state-of-the-art methods on two challenging datasets. The source code is available at https://github.com/Amazingren/PI-Trans.

Yidong Gong, Saima Afrin, Yuchen Ma et al.

Modern deep learning workloads increasingly exhibit dynamic, metadata-driven execution, where runtime-generated information determines memory provisioning and kernel launch decisions. In sampling-based graph neural network (GNN) training, this behavior places the CPU on the critical path, introducing persistent host-device orchestration overhead and frequent GPU-CPU synchronization, which dominate end-to-end runtime when GPU computation is small. Existing approaches, including CUDA Graphs and GPU dynamic parallelism, fail to address this problem because the metadata-driven control loop remains host-mediated, and execution structure varies across iterations. We present ZEROGNN, a system that removes the host from the metadata-driven control loop and enables fully GPU-resident execution under dynamic behavior. ZEROGNN keeps runtime metadata on-device, mediates dynamic execution within a fixed launch structure, and provisions a conservative yet tight execution envelope to restore CUDA Graph replayability. Experiments on sampling-based GNN workloads show that ZEROGNN achieves up to 5.28 x end-to-end speedup, near 100% GPU execution fraction, and memory efficiency comparable to ideal metadata-informed allocation, while enabling strong multi-GPU scaling by eliminating host-side bottlenecks.

Bin Ren, Guofeng Mei, Danda Pani Paudel et al.

Contrastive learning (CL) for Vision Transformers (ViTs) in image domains has achieved performance comparable to CL for traditional convolutional backbones. However, in 3D point cloud pretraining with ViTs, masked autoencoder (MAE) modeling remains dominant. This raises the question: Can we take the best of both worlds? To answer this question, we first empirically validate that integrating MAE-based point cloud pre-training with the standard contrastive learning paradigm, even with meticulous design, can lead to a decrease in performance. To address this limitation, we reintroduce CL into the MAE-based point cloud pre-training paradigm by leveraging the inherent contrastive properties of MAE. Specifically, rather than relying on extensive data augmentation as commonly used in the image domain, we randomly mask the input tokens twice to generate contrastive input pairs. Subsequently, a weight-sharing encoder and two identically structured decoders are utilized to perform masked token reconstruction. Additionally, we propose that for an input token masked by both masks simultaneously, the reconstructed features should be as similar as possible. This naturally establishes an explicit contrastive constraint within the generative MAE-based pre-training paradigm, resulting in our proposed method, Point-CMAE. Consequently, Point-CMAE effectively enhances the representation quality and transfer performance compared to its MAE counterpart. Experimental evaluations across various downstream applications, including classification, part segmentation, and few-shot learning, demonstrate the efficacy of our framework in surpassing state-of-the-art techniques under standard ViTs and single-modal settings. The source code and trained models are available at: https://github.com/Amazingren/Point-CMAE.

Gen Li, Jie Ji, Minghai Qin et al.

As deep convolutional neural networks (DNNs) are widely used in various fields of computer vision, leveraging the overfitting ability of the DNN to achieve video resolution upscaling has become a new trend in the modern video delivery system. By dividing videos into chunks and overfitting each chunk with a super-resolution model, the server encodes videos before transmitting them to the clients, thus achieving better video quality and transmission efficiency. However, a large number of chunks are expected to ensure good overfitting quality, which substantially increases the storage and consumes more bandwidth resources for data transmission. On the other hand, decreasing the number of chunks through training optimization techniques usually requires high model capacity, which significantly slows down execution speed. To reconcile such, we propose a novel method for high-quality and efficient video resolution upscaling tasks, which leverages the spatial-temporal information to accurately divide video into chunks, thus keeping the number of chunks as well as the model size to minimum. Additionally, we advance our method into a single overfitting model by a data-aware joint training technique, which further reduces the storage requirement with negligible quality drop. We deploy our models on an off-the-shelf mobile phone, and experimental results show that our method achieves real-time video super-resolution with high video quality. Compared with the state-of-the-art, our method achieves 28 fps streaming speed with 41.6 PSNR, which is 14$\times$ faster and 2.29 dB better in the live video resolution upscaling tasks. Code available in https://github.com/coulsonlee/STDO-CVPR2023.git

Zifeng Wang, Zheng Zhan, Yifan Gong et al.

Existing work in continual learning (CL) focuses on mitigating catastrophic forgetting, i.e., model performance deterioration on past tasks when learning a new task. However, the training efficiency of a CL system is under-investigated, which limits the real-world application of CL systems under resource-limited scenarios. In this work, we propose a novel framework called Sparse Continual Learning(SparCL), which is the first study that leverages sparsity to enable cost-effective continual learning on edge devices. SparCL achieves both training acceleration and accuracy preservation through the synergy of three aspects: weight sparsity, data efficiency, and gradient sparsity. Specifically, we propose task-aware dynamic masking (TDM) to learn a sparse network throughout the entire CL process, dynamic data removal (DDR) to remove less informative training data, and dynamic gradient masking (DGM) to sparsify the gradient updates. Each of them not only improves efficiency, but also further mitigates catastrophic forgetting. SparCL consistently improves the training efficiency of existing state-of-the-art (SOTA) CL methods by at most 23X less training FLOPs, and, surprisingly, further improves the SOTA accuracy by at most 1.7%. SparCL also outperforms competitive baselines obtained from adapting SOTA sparse training methods to the CL setting in both efficiency and accuracy. We also evaluate the effectiveness of SparCL on a real mobile phone, further indicating the practical potential of our method.

Hongfei Zhang, Harold Haodong Chen, Chenfei Liao et al.

Existing video depth estimation faces a fundamental trade-off: generative models suffer from stochastic geometric hallucinations and scale drift, while discriminative models demand massive labeled datasets to resolve semantic ambiguities. To break this impasse, we present DVD, the first framework to deterministically adapt pre-trained video diffusion models into single-pass depth regressors. Specifically, DVD features three core designs: (i) repurposing the diffusion timestep as a structural anchor to balance global stability with high-frequency details; (ii) latent manifold rectification (LMR) to mitigate regression-induced over-smoothing, enforcing differential constraints to restore sharp boundaries and coherent motion; and (iii) global affine coherence, an inherent property bounding inter-window divergence, which enables seamless long-video inference without requiring complex temporal alignment. Extensive experiments demonstrate that DVD achieves state-of-the-art zero-shot performance across benchmarks. Furthermore, DVD successfully unlocks the profound geometric priors implicit in video foundation models using 163x less task-specific data than leading baselines. Notably, we fully release our pipeline, providing the whole training suite for SOTA video depth estimation to benefit the open-source community.

Qi Ma, Yue Li, Bin Ren et al.

3D Gaussian Splatting (3DGS) has become the de facto method of 3D representation in many vision tasks. This calls for the 3D understanding directly in this representation space. To facilitate the research in this direction, we first build ShapeSplat, a large-scale dataset of 3DGS using the commonly used ShapeNet, ModelNet and Objaverse datasets. Our dataset ShapeSplat consists of 206K objects spanning over 87 unique categories, whose labels are in accordance with the respective datasets. The creation of this dataset utilized the compute equivalent of 3.8 GPU years on a TITAN XP GPU. We utilize our dataset for unsupervised pretraining and supervised finetuning for classification and segmentation tasks. To this end, we introduce Gaussian-MAE, which highlights the unique benefits of representation learning from Gaussian parameters. Through exhaustive experiments, we provide several valuable insights. In particular, we show that (1) the distribution of the optimized GS centroids significantly differs from the uniformly sampled point cloud (used for initialization) counterpart; (2) this change in distribution results in degradation in classification but improvement in segmentation tasks when using only the centroids; (3) to leverage additional Gaussian parameters, we propose Gaussian feature grouping in a normalized feature space, along with splats pooling layer, offering a tailored solution to effectively group and embed similar Gaussians, which leads to notable improvement in finetuning tasks.

Hao Tang, Lei Ding, Songsong Wu et al.

Video processing and analysis have become an urgent task since a huge amount of videos (e.g., Youtube, Hulu) are uploaded online every day. The extraction of representative key frames from videos is very important in video processing and analysis since it greatly reduces computing resources and time. Although great progress has been made recently, large-scale video classification remains an open problem, as the existing methods have not well balanced the performance and efficiency simultaneously. To tackle this problem, this work presents an unsupervised method to retrieve the key frames, which combines Convolutional Neural Network (CNN) and Temporal Segment Density Peaks Clustering (TSDPC). The proposed TSDPC is a generic and powerful framework and it has two advantages compared with previous works, one is that it can calculate the number of key frames automatically. The other is that it can preserve the temporal information of the video. Thus it improves the efficiency of video classification. Furthermore, a Long Short-Term Memory network (LSTM) is added on the top of the CNN to further elevate the performance of classification. Moreover, a weight fusion strategy of different input networks is presented to boost the performance. By optimizing both video classification and key frame extraction simultaneously, we achieve better classification performance and higher efficiency. We evaluate our method on two popular datasets (i.e., HMDB51 and UCF101) and the experimental results consistently demonstrate that our strategy achieves competitive performance and efficiency compared with the state-of-the-art approaches.

Mengyi Zhao, Mengyuan Liu, Bin Ren et al.

Diffusion-based generative models have recently emerged as powerful solutions for high-quality synthesis in multiple domains. Leveraging the bidirectional Markov chains, diffusion probabilistic models generate samples by inferring the reversed Markov chain based on the learned distribution mapping at the forward diffusion process. In this work, we propose Modiff, a conditional paradigm that benefits from the denoising diffusion probabilistic model (DDPM) to tackle the problem of realistic and diverse action-conditioned 3D skeleton-based motion generation. We are a pioneering attempt that uses DDPM to synthesize a variable number of motion sequences conditioned on a categorical action. We evaluate our approach on the large-scale NTU RGB+D dataset and show improvements over state-of-the-art motion generation methods.

Bin Ren, Eduard Zamfir, Zongwei Wu et al.

With the proliferation of mobile devices, the need for an efficient model to restore any degraded image has become increasingly significant and impactful. Traditional approaches typically involve training dedicated models for each specific degradation, resulting in inefficiency and redundancy. More recent solutions either introduce additional modules to learn visual prompts significantly increasing model size or incorporate cross-modal transfer from large language models trained on vast datasets, adding complexity to the system architecture. In contrast, our approach, termed RAM, takes a unified path that leverages inherent similarities across various degradations to enable both efficient and comprehensive restoration through a joint embedding mechanism without scaling up the model or relying on large multimodal models. Specifically, we examine the sub-latent space of each input, identifying key components and reweighting them in a gated manner. This intrinsic degradation awareness is further combined with contextualized attention in an X-shaped framework, enhancing local-global interactions. Extensive benchmarking in an all-in-one restoration setting confirms RAM's SOTA performance, reducing model complexity by approximately 82% in trainable parameters and 85% in FLOPs. Our code and models will be publicly available.

Yushu Wu, Yifan Gong, Pu Zhao et al.

Deep learning-based super-resolution (SR) has gained tremendous popularity in recent years because of its high image quality performance and wide application scenarios. However, prior methods typically suffer from large amounts of computations and huge power consumption, causing difficulties for real-time inference, especially on resource-limited platforms such as mobile devices. To mitigate this, we propose a compiler-aware SR neural architecture search (NAS) framework that conducts depth search and per-layer width search with adaptive SR blocks. The inference speed is directly taken into the optimization along with the SR loss to derive SR models with high image quality while satisfying the real-time inference requirement. Instead of measuring the speed on mobile devices at each iteration during the search process, a speed model incorporated with compiler optimizations is leveraged to predict the inference latency of the SR block with various width configurations for faster convergence. With the proposed framework, we achieve real-time SR inference for implementing 720p resolution with competitive SR performance (in terms of PSNR and SSIM) on GPU/DSP of mobile platforms (Samsung Galaxy S21).

Jou-An Chen, Wei Niu, Bin Ren et al.

Data redundancy is ubiquitous in the inputs and intermediate results of Deep Neural Networks (DNN). It offers many significant opportunities for improving DNN performance and efficiency and has been explored in a large body of work. These studies have scattered in many venues across several years. The targets they focus on range from images to videos and texts, and the techniques they use to detect and exploit data redundancy also vary in many aspects. There is not yet a systematic examination and summary of the many efforts, making it difficult for researchers to get a comprehensive view of the prior work, the state of the art, differences and shared principles, and the areas and directions yet to explore. This article tries to fill the void. It surveys hundreds of recent papers on the topic, introduces a novel taxonomy to put the various techniques into a single categorization framework, offers a comprehensive description of the main methods used for exploiting data redundancy in improving multiple kinds of DNNs on data, and points out a set of research opportunities for future to explore.

Yidi Li, Jiahao Wen, Bin Ren et al.

The integration of point and voxel representations is becoming more common in LiDAR-based 3D object detection. However, this combination often struggles with capturing semantic information effectively. Moreover, relying solely on point features within regions of interest can lead to information loss and limitations in local feature representation. To tackle these challenges, we propose a novel two-stage 3D object detector, called Point-Voxel Attention Fusion Network (PVAFN). PVAFN leverages an attention mechanism to improve multi-modal feature fusion during the feature extraction phase. In the refinement stage, it utilizes a multi-pooling strategy to integrate both multi-scale and region-specific information effectively. The point-voxel attention mechanism adaptively combines point cloud and voxel-based Bird's-Eye-View (BEV) features, resulting in richer object representations that help to reduce false detections. Additionally, a multi-pooling enhancement module is introduced to boost the model's perception capabilities. This module employs cluster pooling and pyramid pooling techniques to efficiently capture key geometric details and fine-grained shape structures, thereby enhancing the integration of local and global features. Extensive experiments on the KITTI and Waymo datasets demonstrate that the proposed PVAFN achieves competitive performance. The code and models will be available.

Yue Li, Qi Ma, Runyi Yang et al.

While 3DGS has emerged as a high-fidelity scene representation, encoding rich, general-purpose features directly from its primitives remains under-explored. We address this gap by introducing Chorus, a multi-teacher pretraining framework that learns a holistic feed-forward 3D Gaussian Splatting (3DGS) scene encoder by distilling complementary signals from 2D foundation models. Chorus employs a shared 3D encoder and teacher-specific projectors to learn from language-aligned, generalist, and object-aware teachers, encouraging a shared embedding space that captures signals from high-level semantics to fine-grained structure. We evaluate Chorus on a wide range of tasks: open-vocabulary semantic and instance segmentation, linear and decoder probing, as well as data-efficient supervision. Besides 3DGS, we also test Chorus on several benchmarks that only support point clouds by pretraining a variant using only Gaussians' centers, colors, estimated normals as inputs. Interestingly, this encoder shows strong transfer and outperforms the point clouds baseline while using 39.9 times fewer training scenes. Finally, we propose a render-and-distill adaptation that facilitates out-of-domain finetuning. Our code and model will be released upon publication.

Jiaze E, Srutarshi Banerjee, Tekin Bicer et al.

Computed tomography (CT) is widely used in scientific imaging systems such as synchrotron and laboratory-based nano-CT, but acquiring full-view sinograms requires high radiation dose and long scan times. Sparse-view CT reduces this burden but produces incomplete sinograms with structured signal loss, degrading reconstruction quality. Unlike RGB images, sinograms encode globally coupled projections and exhibit directional spectral patterns, making conventional RGB-oriented inpainting methods, including diffusion models, ineffective because they ignore angular dependencies and physical constraints inherent to tomographic data. We propose FCDM, a diffusion-based framework for sinogram restoration that incorporates bidirectional frequency reasoning, angular-aware masking, and physics-guided regularization to preserve global structure and physical plausibility. Experiments on real-world datasets show that FCDM consistently outperforms existing baselines, achieving over 0.93 SSIM and 31 dB PSNR across diverse sparse-view settings.

Fulvio Sanguigni, Davide Lobba, Bin Ren et al.

Recent advances in Virtual Try-On (VTON) and Virtual Try-Off (VTOFF) have greatly improved photo-realistic fashion synthesis and garment reconstruction. However, existing datasets remain static, lacking instruction-driven editing for controllable and interactive fashion generation. In this work, we introduce the Dress Editing Dataset (Dress-ED), the first large-scale benchmark that unifies VTON, VTOFF, and text-guided garment editing within a single framework. Each sample in Dress-ED includes an in-shop garment image, the corresponding person image wearing the garment, their edited counterparts, and a natural-language instruction of the desired modification. Built through a fully automated multimodal pipeline that integrates MLLM-based garment understanding, diffusion-based editing, and LLM-guided verification, Dress-ED comprises over 146k verified quadruplets spanning three garment categories and seven edit types, including both appearance (e.g., color, pattern, material) and structural (e.g., sleeve length, neckline) modifications. Based on this benchmark, we further propose a unified multimodal diffusion framework that jointly reasons over linguistic instructions and visual garment cues, serving as a strong baseline for instruction-driven VTON and VTOFF. Dataset and code will be made publicly available.

Bin Ren, Yawei Li, Nancy Mehta et al.

This paper provides a comprehensive review of the NTIRE 2024 challenge, focusing on efficient single-image super-resolution (ESR) solutions and their outcomes. The task of this challenge is to super-resolve an input image with a magnification factor of x4 based on pairs of low and corresponding high-resolution images. The primary objective is to develop networks that optimize various aspects such as runtime, parameters, and FLOPs, while still maintaining a peak signal-to-noise ratio (PSNR) of approximately 26.90 dB on the DIV2K_LSDIR_valid dataset and 26.99 dB on the DIV2K_LSDIR_test dataset. In addition, this challenge has 4 tracks including the main track (overall performance), sub-track 1 (runtime), sub-track 2 (FLOPs), and sub-track 3 (parameters). In the main track, all three metrics (ie runtime, FLOPs, and parameter count) were considered. The ranking of the main track is calculated based on a weighted sum-up of the scores of all other sub-tracks. In sub-track 1, the practical runtime performance of the submissions was evaluated, and the corresponding score was used to determine the ranking. In sub-track 2, the number of FLOPs was considered. The score calculated based on the corresponding FLOPs was used to determine the ranking. In sub-track 3, the number of parameters was considered. The score calculated based on the corresponding parameters was used to determine the ranking. RLFN is set as the baseline for efficiency measurement. The challenge had 262 registered participants, and 34 teams made valid submissions. They gauge the state-of-the-art in efficient single-image super-resolution. To facilitate the reproducibility of the challenge and enable other researchers to build upon these findings, the code and the pre-trained model of validated solutions are made publicly available at https://github.com/Amazingren/NTIRE2024_ESR/.

Yidi Li, Yihan Li, Yixin Guo et al.

In speaker tracking research, integrating and complementing multi-modal data is a crucial strategy for improving the accuracy and robustness of tracking systems. However, tracking with incomplete modalities remains a challenging issue due to noisy observations caused by occlusion, acoustic noise, and sensor failures. Especially when there is missing data in multiple modalities, the performance of existing multi-modal fusion methods tends to decrease. To this end, we propose a Global-Local Distillation-based Tracker (GLDTracker) for robust audio-visual speaker tracking. GLDTracker is driven by a teacher-student distillation model, enabling the flexible fusion of incomplete information from each modality. The teacher network processes global signals captured by camera and microphone arrays, and the student network handles local information subject to visual occlusion and missing audio channels. By transferring knowledge from teacher to student, the student network can better adapt to complex dynamic scenes with incomplete observations. In the student network, a global feature reconstruction module based on the generative adversarial network is constructed to reconstruct global features from feature embedding with missing local information. Furthermore, a multi-modal multi-level fusion attention is introduced to integrate the incomplete feature and the reconstructed feature, leveraging the complementarity and consistency of audio-visual and global-local features. Experimental results on the AV16.3 dataset demonstrate that the proposed GLDTracker outperforms existing state-of-the-art audio-visual trackers and achieves leading performance on both standard and incomplete modalities datasets, highlighting its superiority and robustness in complex conditions. The code and models will be available.

Yuming Huang, Bin Ren, Ziming Xu et al.

Sparse rewards pose a significant challenge to achieving high sample efficiency in goal-conditioned reinforcement learning (RL). Specifically, in sequential manipulation tasks, the agent receives failure rewards until it successfully completes the entire manipulation task, which leads to low sample efficiency. To tackle this issue and improve sample efficiency, we propose a novel model-based RL framework called Model-based Relay Hindsight Experience Replay (MRHER). MRHER breaks down a continuous task into subtasks with increasing complexity and utilizes the previous subtask to guide the learning of the subsequent one. Instead of using Hindsight Experience Replay (HER) in every subtask, we design a new robust model-based relabeling method called Foresight relabeling (FR). FR predicts the future trajectory of the hindsight state and relabels the expected goal as a goal achieved on the virtual future trajectory. By incorporating FR, MRHER effectively captures more information from historical experiences, leading to improved sample efficiency, particularly in object-manipulation environments. Experimental results demonstrate that MRHER exhibits state-of-the-art sample efficiency in benchmark tasks, outperforming RHER by 13.79% and 14.29% in the FetchPush-v1 environment and FetchPickandPlace-v1 environment, respectively.

Mingkang Dong, Hongyi Cai, Jie Li et al.

The effectiveness of multimodal instruction tuning depends not only on dataset scale, but critically on whether training samples genuinely require visual reasoning. However, existing instruction datasets often contain a substantial portion of visually redundant samples (solvable from text alone), as well as multimodally misaligned supervision that can degrade learning. To address this, we propose VisNec (Visual Necessity Score), a principled data selection framework that measures the marginal contribution of visual input during instruction tuning. By comparing predictive loss with and without visual context, VisNec identifies whether a training instance is vision-critical, redundant, or misaligned. To preserve task diversity, we combine VisNec with semantic clustering and select high-necessity samples within each cluster. Across 10 downstream benchmarks, training on only 15% of the LLaVA-665K dataset selected by VisNec achieves 100.2% of full-data performance. On the smaller Vision-Flan-186K dataset, our selection not only further reduces data size but also surpasses full-data training by 15.8%. These results demonstrate that measuring and leveraging visual necessity provides an effective solution for both efficient and robust multimodal instruction tuning. Codes and selected subsets will be released upon acceptance.

Yuqian Fu, Xingyu Qiu, Bin Ren et al.

Cross-Domain Few-Shot Object Detection (CD-FSOD) poses significant challenges to existing object detection and few-shot detection models when applied across domains. In conjunction with NTIRE 2025, we organized the 1st CD-FSOD Challenge, aiming to advance the performance of current object detectors on entirely novel target domains with only limited labeled data. The challenge attracted 152 registered participants, received submissions from 42 teams, and concluded with 13 teams making valid final submissions. Participants approached the task from diverse perspectives, proposing novel models that achieved new state-of-the-art (SOTA) results under both open-source and closed-source settings. In this report, we present an overview of the 1st NTIRE 2025 CD-FSOD Challenge, highlighting the proposed solutions and summarizing the results submitted by the participants.

Yanyu Li, Xuan Shen, Geng Yuan et al.

In this work we demonstrate real-time portrait stylization, specifically, translating self-portrait into cartoon or anime style on mobile devices. We propose a latency-driven differentiable architecture search method, maintaining realistic generative quality. With our framework, we obtain $10\times$ computation reduction on the generative model and achieve real-time video stylization on off-the-shelf smartphone using mobile GPUs.

Xiaofeng Li, Bin Ren, Xipeng Shen et al.

There is a growing demand for shifting the delivery of AI capability from data centers on the cloud to edge or end devices, exemplified by the fast emerging real-time AI-based apps running on smartphones, AR/VR devices, autonomous vehicles, and various IoT devices. The shift has however been seriously hampered by the large growing gap between DNN computing demands and the computing power on edge or end devices. This article presents the design of XGen, an optimizing framework for DNN designed to bridge the gap. XGen takes cross-cutting co-design as its first-order consideration. Its full-stack AI-oriented optimizations consist of a number of innovative optimizations at every layer of the DNN software stack, all designed in a cooperative manner. The unique technology makes XGen able to optimize various DNNs, including those with an extreme depth (e.g., BERT, GPT, other transformers), and generate code that runs several times faster than those from existing DNN frameworks, while delivering the same level of accuracy.

Mengjiao Ma, Qi Ma, Yue Li et al.

3D Gaussian Splatting (3DGS) serves as a highly performant and efficient encoding of scene geometry, appearance, and semantics. Moreover, grounding language in 3D scenes has proven to be an effective strategy for 3D scene understanding. Current Language Gaussian Splatting line of work fall into three main groups: (i) per-scene optimization-based, (ii) per-scene optimization-free, and (iii) generalizable approach. However, most of them are evaluated only on rendered 2D views of a handful of scenes and viewpoints close to the training views, limiting ability and insight into holistic 3D understanding. To address this gap, we propose the first large-scale benchmark that systematically assesses these three groups of methods directly in 3D space, evaluating on 1060 scenes across three indoor datasets and one outdoor dataset. Benchmark results demonstrate a clear advantage of the generalizable paradigm, particularly in relaxing the scene-specific limitation, enabling fast feed-forward inference on novel scenes, and achieving superior segmentation performance. We further introduce GaussianWorld-49K a carefully curated 3DGS dataset comprising around 49K diverse indoor and outdoor scenes obtained from multiple sources, with which we demonstrate the generalizable approach could harness strong data priors. Our codes, benchmark, and datasets will be made public to accelerate research in generalizable 3DGS scene understanding.

Yan Shu, Bin Ren, Zhitong Xiong et al.

Vision-language models (VLMs) have shown promise in earth observation (EO), yet they struggle with tasks that require grounding complex spatial reasoning in precise pixel-level visual representations. To address this problem, we introduce TerraScope, a unified VLM that delivers pixel-grounded geospatial reasoning with two key capabilities: (1) modality-flexible reasoning: it handles single-modality inputs (optical or SAR) and adaptively fuses different modalities into the reasoning process when both are available; (2) multi-temporal reasoning: it integrates temporal sequences for change analysis across multiple time points. In addition, we curate Terra-CoT, a large-scale dataset containing 1 million samples with pixel-level masks embedded in reasoning chains across multiple sources. We also propose TerraScope-Bench, the first benchmark for pixel-grounded geospatial reasoning with six sub-tasks that evaluates both answer accuracy and mask quality to ensure authentic pixel-grounded reasoning. Experiments show that TerraScope significantly outperforms existing VLMs on pixel-grounded geospatial reasoning while providing interpretable visual evidence.

Bin Ren, Eduard Zamfir, Zongwei Wu et al.

Restoring any degraded image efficiently via just one model has become increasingly significant and impactful, especially with the proliferation of mobile devices. Traditional solutions typically involve training dedicated models per degradation, resulting in inefficiency and redundancy. More recent approaches either introduce additional modules to learn visual prompts, significantly increasing model size, or incorporate cross-modal transfer from large language models trained on vast datasets, adding complexity to the system architecture. In contrast, our approach, termed AnyIR, takes a unified path that leverages inherent similarity across various degradations to enable both efficient and comprehensive restoration through a joint embedding mechanism, without scaling up the model or relying on large language models.Specifically, we examine the sub-latent space of each input, identifying key components and reweighting them first in a gated manner. To fuse the intrinsic degradation awareness and the contextualized attention, a spatial-frequency parallel fusion strategy is proposed for enhancing spatial-aware local-global interactions and enriching the restoration details from the frequency perspective. Extensive benchmarking in the all-in-one restoration setting confirms AnyIR's SOTA performance, reducing model complexity by around 82\% in parameters and 85\% in FLOPs. Our code will be available at our Project page (https://amazingren.github.io/AnyIR/)

Zixin Zhang, Chenfei Liao, Hongfei Zhang et al.

Affordance prediction serves as a critical bridge between perception and action in embodied AI. However, existing research is confined to pinhole camera models, which suffer from narrow Fields of View (FoV) and fragmented observations, often missing critical holistic environmental context. In this paper, we present the first exploration into Panoramic Affordance Prediction, utilizing 360-degree imagery to capture global spatial relationships and holistic scene understanding. To facilitate this novel task, we first introduce PAP-12K, a large-scale benchmark dataset containing over 1,000 ultra-high-resolution (12k, 11904 x 5952) panoramic images with over 12k carefully annotated QA pairs and affordance masks. Furthermore, we propose PAP, a training-free, coarse-to-fine pipeline inspired by the human foveal visual system to tackle the ultra-high resolution and severe distortion inherent in panoramic images. PAP employs recursive visual routing via grid prompting to progressively locate targets, applies an adaptive gaze mechanism to rectify local geometric distortions, and utilizes a cascaded grounding pipeline to extract precise instance-level masks. Experimental results on PAP-12K reveal that existing affordance prediction methods designed for standard perspective images suffer severe performance degradation and fail due to the unique challenges of panoramic vision. In contrast, PAP framework effectively overcomes these obstacles, significantly outperforming state-of-the-art baselines and highlighting the immense potential of panoramic perception for robust embodied intelligence.

Xu Zheng, Zihao Dongfang, Lutao Jiang et al.

Humans possess spatial reasoning abilities that enable them to understand spaces through multimodal observations, such as vision and sound. Large multimodal reasoning models extend these abilities by learning to perceive and reason, showing promising performance across diverse spatial tasks. However, systematic reviews and publicly available benchmarks for these models remain limited. In this survey, we provide a comprehensive review of multimodal spatial reasoning tasks with large models, categorizing recent progress in multimodal large language models (MLLMs) and introducing open benchmarks for evaluation. We begin by outlining general spatial reasoning, focusing on post-training techniques, explainability, and architecture. Beyond classical 2D tasks, we examine spatial relationship reasoning, scene and layout understanding, as well as visual question answering and grounding in 3D space. We also review advances in embodied AI, including vision-language navigation and action models. Additionally, we consider emerging modalities such as audio and egocentric video, which contribute to novel spatial understanding through new sensors. We believe this survey establishes a solid foundation and offers insights into the growing field of multimodal spatial reasoning. Updated information about this survey, codes and implementation of the open benchmarks can be found at https://github.com/zhengxuJosh/Awesome-Spatial-Reasoning.

Bin Ren, Yawei Li, Xu Zheng et al.

Image Restoration (IR) aims to recover high quality images from degraded inputs affected by various corruptions such as noise, blur, haze, rain, and low light conditions. Despite recent advances, most existing approaches treat IR as a direct mapping problem, relying on shared representations across degradation types without modeling their structural diversity. In this work, we present MIRAGE, a unified and lightweight framework for all in one IR that explicitly decomposes the input feature space into three semantically aligned parallel branches, each processed by a specialized module attention for global context, convolution for local textures, and MLP for channel-wise statistics. This modular decomposition significantly improves generalization and efficiency across diverse degradations. Furthermore, we introduce a cross layer contrastive learning scheme that aligns shallow and latent features to enhance the discriminability of shared representations. To better capture the underlying geometry of feature representations, we perform contrastive learning in a Symmetric Positive Definite (SPD) manifold space rather than the conventional Euclidean space. Extensive experiments show that MIRAGE not only achieves new state of the art performance across a variety of degradation types but also offers a scalable solution for challenging all-in-one IR scenarios. Our code and models will be publicly available at https://amazingren.github.io/MIRAGE/.

Ti Wang, Mengyuan Liu, Hong Liu et al.

Although data-driven methods have achieved success in 3D human pose estimation, they often suffer from domain gaps and exhibit limited generalization. In contrast, optimization-based methods excel in fine-tuning for specific cases but are generally inferior to data-driven methods in overall performance. We observe that previous optimization-based methods commonly rely on a projection constraint, which only ensures alignment in 2D space, potentially leading to the overfitting problem. To address this, we propose an Uncertainty-Aware testing-time Optimization (UAO) framework, which keeps the prior information of the pre-trained model and alleviates the overfitting problem using the uncertainty of joints. Specifically, during the training phase, we design an effective 2D-to-3D network for estimating the corresponding 3D pose while quantifying the uncertainty of each 3D joint. For optimization during testing, the proposed optimization framework freezes the pre-trained model and optimizes only a latent state. Projection loss is then employed to ensure the generated poses are well aligned in 2D space for high-quality optimization. Furthermore, we utilize the uncertainty of each joint to determine how much each joint is allowed for optimization. The effectiveness and superiority of the proposed framework are validated through extensive experiments on challenging datasets: Human3.6M, MPI-INF-3DHP, and 3DPW. Notably, our approach outperforms the previous best result by a large margin of 5.5\% on Human3.6M. Code is available at \href{https://github.com/xiu-cs/UAO-Pose3D}{https://github.com/xiu-cs/UAO-Pose3D}.

Chenfei Liao, Wensong Wang, Zichen Wen et al.

Recent endeavors to accelerate inference in Multimodal Large Language Models (MLLMs) have primarily focused on visual token compression. The effectiveness of these methods is typically assessed by measuring the accuracy drop on established benchmarks, comparing model performance before and after compression. However, these benchmarks are originally designed to assess the perception and reasoning capabilities of MLLMs, rather than to evaluate compression techniques. As a result, directly applying them to visual token compression introduces a task mismatch. Strikingly, our investigation reveals that simple image downsampling consistently outperforms many advanced compression methods across multiple widely used benchmarks. Through extensive experiments, we make the following observations: (i) Current benchmarks are noisy for the visual token compression task. (ii) Down-sampling is able to serve as a data filter to evaluate the difficulty of samples in the visual token compression task. Motivated by these findings, we introduce VTC-Bench, an evaluation framework that incorporates a data filtering mechanism to denoise existing benchmarks, thereby enabling fairer and more accurate assessment of visual token compression methods. All data and code are available at https://github.com/Chenfei-Liao/VTC-Bench.

Yihong Wu, Jinqiao Wei, Xionghui Zhao et al.

Deep learning-based crowd counting methods have achieved remarkable progress in recent years. However, in complex crowd scenarios, existing models still face challenges when adapting to significant density distribution differences between regions. Additionally, the inconsistency of individual representations caused by viewpoint changes and body posture differences further limits the counting accuracy of the models. To address these challenges, we propose DSGC-Net, a Dual-Stream Graph Convolutional Network based on feature correlation mining. DSGC-Net introduces a Density Approximation (DA) branch and a Representation Approximation (RA) branch. By modeling two semantic graphs, it captures the potential feature correlations in density variations and representation distributions. The DA branch incorporates a density prediction module that generates the density distribution map, and constructs a density-driven semantic graph based on density similarity. The RA branch establishes a representation-driven semantic graph by computing global representation similarity. Then, graph convolutional networks are applied to the two semantic graphs separately to model the latent semantic relationships, which enhance the model's ability to adapt to density variations and improve counting accuracy in multi-view and multi-pose scenarios. Extensive experiments on three widely used datasets demonstrate that DSGC-Net outperforms current state-of-the-art methods. In particular, we achieve MAE of 48.9 and 5.9 in ShanghaiTech Part A and Part B datasets, respectively. The released code is available at: https://github.com/Wu-eon/CrowdCounting-DSGCNet.

Bin Ren, Xiaoshui Huang, Mengyuan Liu et al.

Vision transformers (ViTs) have recently been widely applied to 3D point cloud understanding, with masked autoencoding as the predominant pre-training paradigm. However, the challenge of learning dense and informative semantic features from point clouds via standard ViTs remains underexplored. We propose MaskClu, a novel unsupervised pre-training method for ViTs on 3D point clouds that integrates masked point modeling with clustering-based learning. MaskClu is designed to reconstruct both cluster assignments and cluster centers from masked point clouds, thus encouraging the model to capture dense semantic information. Additionally, we introduce a global contrastive learning mechanism that enhances instance-level feature learning by contrasting different masked views of the same point cloud. By jointly optimizing these complementary objectives, i.e., dense semantic reconstruction, and instance-level contrastive learning. MaskClu enables ViTs to learn richer and more semantically meaningful representations from 3D point clouds. We validate the effectiveness of our method via multiple 3D tasks, including part segmentation, semantic segmentation, object detection, and classification, where MaskClu sets new competitive results. The code and models will be released at:https://github.com/Amazingren/maskclu.

Geng Yuan, Xiaolong Ma, Wei Niu et al.

Recently, a new trend of exploring sparsity for accelerating neural network training has emerged, embracing the paradigm of training on the edge. This paper proposes a novel Memory-Economic Sparse Training (MEST) framework targeting for accurate and fast execution on edge devices. The proposed MEST framework consists of enhancements by Elastic Mutation (EM) and Soft Memory Bound (&S) that ensure superior accuracy at high sparsity ratios. Different from the existing works for sparse training, this current work reveals the importance of sparsity schemes on the performance of sparse training in terms of accuracy as well as training speed on real edge devices. On top of that, the paper proposes to employ data efficiency for further acceleration of sparse training. Our results suggest that unforgettable examples can be identified in-situ even during the dynamic exploration of sparsity masks in the sparse training process, and therefore can be removed for further training speedup on edge devices. Comparing with state-of-the-art (SOTA) works on accuracy, our MEST increases Top-1 accuracy significantly on ImageNet when using the same unstructured sparsity scheme. Systematical evaluation on accuracy, training speed, and memory footprint are conducted, where the proposed MEST framework consistently outperforms representative SOTA works. A reviewer strongly against our work based on his false assumptions and misunderstandings. On top of the previous submission, we employ data efficiency for further acceleration of sparse training. And we explore the impact of model sparsity, sparsity schemes, and sparse training algorithms on the number of removable training examples. Our codes are publicly available at: https://github.com/boone891214/MEST.

Bin Ren, Hao Tang, Nicu Sebe

It is hard to generate an image at target view well for previous cross-view image translation methods that directly adopt a simple encoder-decoder or U-Net structure, especially for drastically different views and severe deformation cases. To ease this problem, we propose a novel two-stage framework with a new Cascaded Cross MLP-Mixer (CrossMLP) sub-network in the first stage and one refined pixel-level loss in the second stage. In the first stage, the CrossMLP sub-network learns the latent transformation cues between image code and semantic map code via our novel CrossMLP blocks. Then the coarse results are generated progressively under the guidance of those cues. Moreover, in the second stage, we design a refined pixel-level loss that eases the noisy semantic label problem with more reasonable regularization in a more compact fashion for better optimization. Extensive experimental results on Dayton~\cite{vo2016localizing} and CVUSA~\cite{workman2015wide} datasets show that our method can generate significantly better results than state-of-the-art methods. The source code and trained models are available at https://github.com/Amazingren/CrossMLP.

Yuxuan Cai, Hongjia Li, Geng Yuan et al.

The rapid development and wide utilization of object detection techniques have aroused attention on both accuracy and speed of object detectors. However, the current state-of-the-art object detection works are either accuracy-oriented using a large model but leading to high latency or speed-oriented using a lightweight model but sacrificing accuracy. In this work, we propose YOLObile framework, a real-time object detection on mobile devices via compression-compilation co-design. A novel block-punched pruning scheme is proposed for any kernel size. To improve computational efficiency on mobile devices, a GPU-CPU collaborative scheme is adopted along with advanced compiler-assisted optimizations. Experimental results indicate that our pruning scheme achieves 14$\times$ compression rate of YOLOv4 with 49.0 mAP. Under our YOLObile framework, we achieve 17 FPS inference speed using GPU on Samsung Galaxy S20. By incorporating our proposed GPU-CPU collaborative scheme, the inference speed is increased to 19.1 FPS, and outperforms the original YOLOv4 by 5$\times$ speedup. Source code is at: \url{https://github.com/nightsnack/YOLObile}.

Bin Ren, Hang Guo, Yan Shu et al.

This paper reviews the NTIRE 2026 challenge on efficient single-image super-resolution with a focus on the proposed solutions and results. The aim of this challenge is to devise a network that reduces one or several aspects, such as runtime, parameters, and FLOPs, while maintaining PSNR of around 26.90 dB on the DIV2K_LSDIR_valid dataset, and 26.99 dB on the DIV2K_LSDIR_test dataset. The challenge had 95 registered participants, and 15 teams made valid submissions. They gauge the state-of-the-art results for efficient single-image super-resolution.

Bin Ren, Hang Guo, Lei Sun et al.

This paper presents a comprehensive review of the NTIRE 2025 Challenge on Single-Image Efficient Super-Resolution (ESR). The challenge aimed to advance the development of deep models that optimize key computational metrics, i.e., runtime, parameters, and FLOPs, while achieving a PSNR of at least 26.90 dB on the $\operatorname{DIV2K\_LSDIR\_valid}$ dataset and 26.99 dB on the $\operatorname{DIV2K\_LSDIR\_test}$ dataset. A robust participation saw \textbf{244} registered entrants, with \textbf{43} teams submitting valid entries. This report meticulously analyzes these methods and results, emphasizing groundbreaking advancements in state-of-the-art single-image ESR techniques. The analysis highlights innovative approaches and establishes benchmarks for future research in the field.

Xiaoning Liu, Zongwei Wu, Florin-Alexandru Vasluianu et al.

This paper presents a comprehensive review of the NTIRE 2025 Low-Light Image Enhancement (LLIE) Challenge, highlighting the proposed solutions and final outcomes. The objective of the challenge is to identify effective networks capable of producing brighter, clearer, and visually compelling images under diverse and challenging conditions. A remarkable total of 762 participants registered for the competition, with 28 teams ultimately submitting valid entries. This paper thoroughly evaluates the state-of-the-art advancements in LLIE, showcasing the significant progress.

Shuhong Liu, Chenyu Bao, Ziteng Cui et al.

This paper presents a comprehensive review of the NTIRE 2026 3D Restoration and Reconstruction (3DRR) Challenge, detailing the proposed methods and results. The challenge seeks to identify robust reconstruction pipelines that are robust under real-world adverse conditions, specifically extreme low-light and smoke-degraded environments, as captured by our RealX3D benchmark. A total of 279 participants registered for the competition, of whom 33 teams submitted valid results. We thoroughly evaluate the submitted approaches against state-of-the-art baselines, revealing significant progress in 3D reconstruction under adverse conditions. Our analysis highlights shared design principles among top-performing methods and provides insights into effective strategies for handling 3D scene degradation.

Lei Sun, Hang Guo, Bin Ren et al.

This paper presents an overview of the NTIRE 2025 Image Denoising Challenge (σ = 50), highlighting the proposed methodologies and corresponding results. The primary objective is to develop a network architecture capable of achieving high-quality denoising performance, quantitatively evaluated using PSNR, without constraints on computational complexity or model size. The task assumes independent additive white Gaussian noise (AWGN) with a fixed noise level of 50. A total of 290 participants registered for the challenge, with 20 teams successfully submitting valid results, providing insights into the current state-of-the-art in image denoising.

Xu Zheng, Ziqiao Weng, Yuanhuiyi Lyu et al.

Retrieval-augmented generation (RAG) has emerged as a pivotal technique in artificial intelligence (AI), particularly in enhancing the capabilities of large language models (LLMs) by enabling access to external, reliable, and up-to-date knowledge sources. In the context of AI-Generated Content (AIGC), RAG has proven invaluable by augmenting model outputs with supplementary, relevant information, thus improving their quality. Recently, the potential of RAG has extended beyond natural language processing, with emerging methods integrating retrieval-augmented strategies into the computer vision (CV) domain. These approaches aim to address the limitations of relying solely on internal model knowledge by incorporating authoritative external knowledge bases, thereby improving both the understanding and generation capabilities of vision models. This survey provides a comprehensive review of the current state of retrieval-augmented techniques in CV, focusing on two main areas: (I) visual understanding and (II) visual generation. In the realm of visual understanding, we systematically review tasks ranging from basic image recognition to complex applications such as medical report generation and multimodal question answering. For visual content generation, we examine the application of RAG in tasks related to image, video, and 3D generation. Furthermore, we explore recent advancements in RAG for embodied AI, with a particular focus on applications in planning, task execution, multimodal perception, interaction, and specialized domains. Given that the integration of retrieval-augmented techniques in CV is still in its early stages, we also highlight the key limitations of current approaches and propose future research directions to drive the development of this promising area.

Yue Li, Qi Ma, Runyi Yang et al.

Recognizing arbitrary or previously unseen categories is essential for comprehensive real-world 3D scene understanding. Currently, all existing methods rely on 2D or textual modalities during training or together at inference. This highlights the clear absence of a model capable of processing 3D data alone for learning semantics end-to-end, along with the necessary data to train such a model. Meanwhile, 3D Gaussian Splatting (3DGS) has emerged as the de facto standard for 3D scene representation across various vision tasks. However, effectively integrating semantic reasoning into 3DGS in a generalizable manner remains an open challenge. To address these limitations, we introduce SceneSplat, to our knowledge the first large-scale 3D indoor scene understanding approach that operates natively on 3DGS. Furthermore, we propose a self-supervised learning scheme that unlocks rich 3D feature learning from unlabeled scenes. To power the proposed methods, we introduce SceneSplat-7K, the first large-scale 3DGS dataset for indoor scenes, comprising 7916 scenes derived from seven established datasets, such as ScanNet and Matterport3D. Generating SceneSplat-7K required computational resources equivalent to 150 GPU days on an L4 GPU, enabling standardized benchmarking for 3DGS-based reasoning for indoor scenes. Our exhaustive experiments on SceneSplat-7K demonstrate the significant benefit of the proposed method over the established baselines.

Xiang Chen, Hao Li, Jiangxin Dong et al.

This paper presents a review for the LoViF Challenge on Real-World All-in-One Image Restoration. The challenge aimed to advance research on real-world all-in-one image restoration under diverse real-world degradation conditions, including blur, low-light, haze, rain, and snow. It provided a unified benchmark to evaluate the robustness and generalization ability of restoration models across multiple degradation categories within a common framework. The competition attracted 124 registered participants and received 9 valid final submissions with corresponding fact sheets, significantly contributing to the progress of real-world all-in-one image restoration. This report provides a detailed analysis of the submitted methods and corresponding results, emphasizing recent progress in unified real-world image restoration. The analysis highlights effective approaches and establishes a benchmark for future research in real-world low-level vision.

Wei Niu, Md Musfiqur Rahman Sanim, Zhihao Shu et al.

This work is motivated by recent developments in Deep Neural Networks, particularly the Transformer architectures underlying applications such as ChatGPT, and the need for performing inference on mobile devices. Focusing on emerging transformers (specifically the ones with computationally efficient Swin-like architectures) and large models (e.g., Stable Diffusion and LLMs) based on transformers, we observe that layout transformations between the computational operators cause a significant slowdown in these applications. This paper presents SmartMem, a comprehensive framework for eliminating most layout transformations, with the idea that multiple operators can use the same tensor layout through careful choice of layout and implementation of operations. Our approach is based on classifying the operators into four groups, and considering combinations of producer-consumer edges between the operators. We develop a set of methods for searching such layouts. Another component of our work is developing efficient memory layouts for 2.5 dimensional memory commonly seen in mobile devices. Our experimental results show that SmartMem outperforms 5 state-of-the-art DNN execution frameworks on mobile devices across 18 varied neural networks, including CNNs, Transformers with both local and global attention, as well as LLMs. In particular, compared to DNNFusion, SmartMem achieves an average speedup of 2.8$\times$, and outperforms TVM and MNN with speedups of 6.9$\times$ and 7.9$\times$, respectively, on average.

Xu Zheng, Chenfei Liao, Yuqian Fu et al.

Recent advances in Multimodal Large Language Models (MLLMs) have shown promising results in integrating diverse modalities such as texts and images. MLLMs are heavily influenced by modality bias, often relying on language while under-utilizing other modalities like visual inputs. This position paper argues that MLLMs are deeply affected by modality bias. Firstly, we diagnose the current state of modality bias, highlighting its manifestations across various tasks. Secondly, we propose a systematic research road-map related to modality bias in MLLMs. Thirdly, we identify key factors of modality bias in MLLMs and offer actionable suggestions for future research to mitigate it. To substantiate these findings, we conduct experiments that demonstrate the influence of each factor: 1. Data Characteristics: Language data is compact and abstract, while visual data is redundant and complex, creating an inherent imbalance in learning dynamics. 2. Imbalanced Backbone Capabilities: The dominance of pretrained language models in MLLMs leads to overreliance on language and neglect of visual information. 3. Training Objectives: Current objectives often fail to promote balanced cross-modal alignment, resulting in shortcut learning biased toward language. These findings highlight the need for balanced training strategies and model architectures to better integrate multiple modalities in MLLMs. We call for interdisciplinary efforts to tackle these challenges and drive innovation in MLLM research. Our work provides a fresh perspective on modality bias in MLLMs and offers insights for developing more robust and generalizable multimodal systems-advancing progress toward Artificial General Intelligence.

Yinchuan Wang, Bin Ren, Xiang Zhang et al.

LiDAR-based SLAM is recognized as one effective method to offer localization guidance in rough environments. However, off-the-shelf LiDAR-based SLAM methods suffer from significant pose estimation drifts, particularly components relevant to the vertical direction, when passing to uneven terrains. This deficiency typically leads to a conspicuously distorted global map. In this article, a LiDAR-based SLAM method is presented to improve the accuracy of pose estimations for ground vehicles in rough terrains, which is termed Rotation-Optimized LiDAR-Only (ROLO) SLAM. The method exploits a forward location prediction to coarsely eliminate the location difference of consecutive scans, thereby enabling separate and accurate determination of the location and orientation at the front-end. Furthermore, we adopt a parallel-capable spatial voxelization for correspondence-matching. We develop a spherical alignment-guided rotation registration within each voxel to estimate the rotation of vehicle. By incorporating geometric alignment, we introduce the motion constraint into the optimization formulation to enhance the rapid and effective estimation of LiDAR's translation. Subsequently, we extract several keyframes to construct the submap and exploit an alignment from the current scan to the submap for precise pose estimation. Meanwhile, a global-scale factor graph is established to aid in the reduction of cumulative errors. In various scenes, diverse experiments have been conducted to evaluate our method. The results demonstrate that ROLO-SLAM excels in pose estimation of ground vehicles and outperforms existing state-of-the-art LiDAR SLAM frameworks.

Wei Niu, Gagan Agrawal, Bin Ren

Though many compilation and runtime systems have been developed for DNNs in recent years, the focus has largely been on static DNNs. Dynamic DNNs, where tensor shapes and sizes and even the set of operators used are dependent upon the input and/or execution, are becoming common. This paper presents SoD$^2$, a comprehensive framework for optimizing Dynamic DNNs. The basis of our approach is a classification of common operators that form DNNs, and the use of this classification towards a Rank and Dimension Propagation (RDP) method. This framework statically determines the shapes of operators as known constants, symbolic constants, or operations on these. Next, using RDP we enable a series of optimizations, like fused code generation, execution (order) planning, and even runtime memory allocation plan generation. By evaluating the framework on 10 emerging Dynamic DNNs and comparing it against several existing systems, we demonstrate both reductions in execution latency and memory requirements, with RDP-enabled key optimizations responsible for much of the gains. Our evaluation results show that SoD$^2$ runs up to $3.9\times$ faster than these systems while saving up to $88\%$ peak memory consumption.