In this study, we introduce EinsPT, an efficient instance-aware pre-training paradigm designed to reduce the transfer gap between vision foundation models and downstream instance-level tasks. Unlike conventional image-level pre-training that relies solely on unlabeled images, EinsPT leverages both image reconstruction and instance annotations to learn representations that are spatially coherent and instance discriminative. To achieve this efficiently, we propose a proxy-foundation architecture that decouples high-resolution and low-resolution learning: the foundation model processes masked low-resolution images for global semantics, while a lightweight proxy model operates on complete high-resolution images to preserve fine-grained details. The two branches are jointly optimized through reconstruction and instance-level prediction losses on fused features. Extensive experiments demonstrate that EinsPT consistently enhances recognition accuracy across various downstream tasks with substantially reduced computational cost, while qualitative results further reveal improved instance perception and completeness in visual representations. Code is available at github.com/feufhd/EinsPT.

In this study, we establish a benchmark and a baseline approach for Multimodal referring and grounding with Chain-of-Questions (MCQ), opening up a promising direction for ‘logical’ multimodal dialogues. The newly collected dataset, named CB-300K, spans challenges including probing dialogues with spatial relationship among multiple objects, consistent reasoning, and complex question chains. The baseline approach, termed ChatterBox, involves a modularized design and a referent feedback mechanism to ensure logical coherence in continuous referring and grounding tasks. This design reduces the risk of referential confusion, simplifies the training process, and presents validity in retaining the language model’s generation ability. Experiments show that ChatterBox demonstrates superiority in MCQ both quantitatively and qualitatively, paving a new path towards multimodal dialogue scenarios with logical interactions.

Multimodal large language models (MLLMs) have enabled open-world visual understanding by injecting visual input as extra tokens into large language models (LLMs) as contexts. However, when the visual input changes from a single image to a long video, the above paradigm encounters difficulty because the vast amount of video tokens has significantly exceeded the maximal capacity of MLLMs. Therefore, existing video-based MLLMs are mostly established upon sampling a small portion of tokens from input data, which can cause key information to be lost and thus produce incorrect answers. This paper presents a simple yet effective algorithm named Adaptive Keyframe Sampling (AKS). It inserts a plug-and-play module known as keyframe selection, which aims to maximize the useful information with a fixed number of video tokens. We formulate keyframe selection as an optimization involving (1) the relevance between the keyframes and the prompt, and (2) the coverage of the keyframes over the video, and present an adaptive algorithm to approximate the best solution. Experiments on two long video understanding benchmarks validate that AKS improves video QA accuracy (beyond strong baselines) upon selecting informative keyframes. Our study reveals the importance of information pre-filtering in video-based MLLMs. Our codes are available at https://github.com/ncTimTang/AKS

Visual representation models leveraging attention mechanisms are challenged by significant computational overhead, particularly when pursuing large receptive fields. In this study, we aim to mitigate this challenge by introducing the Heat Conduction Operator (HCO) built upon the physical heat conduction principle. HCO conceptualizes image patches as heat sources and models their correlations through adaptive thermal energy diffusion, enabling robust visual representations. HCO enjoys a computational complexity of O(N1.5), as it can be implemented using discrete cosine transformation (DCT) operations. HCO is plug-and-play, combining with deep learning backbones produces visual representation models (termed vHeat) with global receptive fields. Experiments across vision tasks demonstrate that, beyond the stronger performance, vHeat achieves up to a 3× throughput, 80% less GPU memory allocation, and 35% fewer computational FLOPs compared to the Swin-Transformer. Code is available at https://github.com/MzeroMiko/vHeat and https://openi.pcl.ac.cn/georgew/vHeat.

Aligning vision and language concepts at a finer level remains an essential topic of multimodal large language models (MLLMs), particularly for tasks such as referring and grounding. Existing methods, such as *proxy encoding* and *geometry encoding* genres, incorporate additional syntax to encode spatial information, imposing extra burdens when communicating between language with vision modules. In this study, we propose ClawMachine, offering a new methodology that explicitly notates each entity using **token collectives**—groups of visual tokens that collaboratively represent higher-level semantics. A hybrid perception mechanism is also explored to perceive and understand scenes from both discrete and continuous spaces. Our method unifies the prompt and answer of visual referential tasks without using additional syntax. By leveraging a joint vision-language vocabulary, ClawMachine integrates referring and grounding in an auto-regressive manner, demonstrating great potential with scaled up pre-training data. Experiments show that ClawMachine achieves superior performance on scene-level and referential understanding tasks with higher efficiency. It also exhibits the potential to integrate multi-source information for complex visual reasoning, which is beyond the capability of many MLLMs. Our code is available at https://github.com/martian422/ClawMachine.

Recent advances in diffusion models have revolutionized text-guided image editing, yet existing editing methods face critical challenges in hyperparameter identification. To get the reasonable editing performance, these methods often require the user to brute-force tune multiple interdependent hyperparameters, such as inversion timesteps and attention modification, \textit{etc.} This process incurs high computational costs due to the huge hyperparameter search space. We consider searching optimal editing's hyperparameters as a sequential decision-making task within the diffusion denoising process. Specifically, we propose a reinforcement learning framework, which establishes a Markov Decision Process that dynamically adjusts hyperparameters across denoising steps, integrating editing objectives into a reward function. The method achieves time efficiency through proximal policy optimization while maintaining optimal hyperparameter configurations. Experiments demonstrate significant reduction in search time and computational overhead compared to existing brute-force approaches, advancing the practical deployment of a diffusion-based image editing framework in the real world.

Enhancing the network architecture of the YOLO framework has been crucial for a long time. Still, it has focused on CNN-based improvements despite the proven superiority of attention mechanisms in modeling capabilities. This is because attention-based models cannot match the speed of CNN-based models. This paper proposes an attention-centric YOLO framework, namely YOLOv12, that matches the speed of previous CNN-based ones while harnessing the performance benefits of attention mechanisms. YOLOv12 surpasses popular real-time object detectors in accuracy with competitive speed. For example, YOLOv12-N achieves 40.5% mAP with an inference latency of 1.62 ms on a T4 GPU, outperforming advanced YOLOv10-N / YOLO11-N by 2.0%/1.1% mAP with a comparable speed. This advantage extends to other model scales. YOLOv12 also surpasses end-to-end real-time detectors that improve DETR, such as RT-DETRv2 / RT-DETRv3: YOLOv12-X beats RT-DETRv2-R101 / RT-DETRv3-R101 while running faster with fewer computations and parameters. See more comparisons in Figure 1. Source code is available at https://github.com/sunsmarterjie/yolov12.

Conventional neural architecture search (NAS) algorithms typically work on search spaces with short-distance node connections. We argue that such designs, though safe and stable, are obstacles to exploring more effective network architectures. In this brief, we explore the search algorithm upon a complicated search space with long-distance connections and show that existing weight-sharing search algorithms fail due to the existence of interleaved connections (ICs). Based on the observation, we present a simple-yet-effective algorithm, termed interleaving-free neural architecture search (IF-NAS). We further design a periodic sampling strategy to construct subnetworks during the search procedure, avoiding the ICs to emerge in any of them. In the proposed search space, IF-NAS outperforms both random sampling and previous weight-sharing search algorithms by significant margins. It can also be well-generalized to the microcell-based spaces. This study emphasizes the importance of macrostructure and we look forward to further efforts in this direction. The code is available at github.com/sunsmarterjie/IFNAS.

Vision Transformers (ViTs) have achieved remarkable success in computer vision tasks. However, their potential in rotation-sensitive scenarios has not been fully explored, and this limitation may be inherently attributed to the lack of spatial invariance in the data-forwarding process. In this study, we present a novel approach, termed Spatial Transform Decoupling (STD), providing a simple-yet-effective solution for oriented object detection with ViTs. Built upon stacked ViT blocks, STD utilizes separate network branches to predict the position, size, and angle of bounding boxes, effectively harnessing the spatial transform potential of ViTs in a divide-and-conquer fashion. Moreover, by aggregating cascaded activation masks (CAMs) computed upon the regressed parameters, STD gradually enhances features within regions of interest (RoIs), which complements the self-attention mechanism. Without bells and whistles, STD achieves state-of-the-art performance on the benchmark datasets including DOTA-v1.0 (82.24% mAP) and HRSC2016 (98.55% mAP), which demonstrates the effectiveness of the proposed method. Source code is available at https://github.com/yuhongtian17/Spatial-Transform-Decoupling.

Designing computationally efficient network architectures remains an ongoing necessity in computer vision. In this paper, we adapt Mamba, a state-space language model, into VMamba, a vision backbone with linear time complexity. At the core of VMamba is a stack of Visual State-Space (VSS) blocks with the 2D Selective Scan (SS2D) module. By traversing along four scanning routes, SS2D bridges the gap between the ordered nature of 1D selective scan and the non-sequential structure of 2D vision data, which facilitates the collection of contextual information from various sources and perspectives. Based on the VSS blocks, we develop a family of VMamba architectures and accelerate them through a succession of architectural and implementation enhancements. Extensive experiments demonstrate VMamba’s promising performance across diverse visual perception tasks, highlighting its superior input scaling efficiency compared to existing benchmark models. Source code is available at https://github.com/MzeroMiko/VMamba

Videos carry rich visual information including object description, action, interaction, etc., but the existing multimodal large language models (MLLMs) fell short in referential understanding scenarios such as video-based referring. In this paper, we present Artemis, an MLLM that pushes video-based referential understanding to a finer level. Given a video, Artemis receives a natural-language question with a bounding box in any video frame and describes the referred target in the entire video. The key to achieving this goal lies in extracting compact, target-specific video features, where we set a solid baseline by tracking and selecting spatiotemporal features from the video. We train Artemis on the newly established ViderRef45K dataset with 45K video-QA pairs and design a computationally efficient, three-stage training procedure. Results are promising both quantitatively and qualitatively. Additionally, we show that Artemis can be integrated with video grounding and text summarization tools to understand more complex scenarios. Code and data are available at https://github.com/NeurIPS24Artemis/Artemis.

We propose integrally pre-trained transformer pyramid network (iTPN), towards jointly optimizing the network backbone and the neck, so that transfer gap between representation models and downstream tasks is minimal. iTPN is born with two elaborated designs: 1) The first pre-trained feature pyramid upon vision transformer (ViT). 2) Multi-stage supervision to the feature pyramid using masked feature modeling (MFM). iTPN is updated to Fast-iTPN, reducing computational memory overhead and accelerating inference through two flexible designs. 1) Token migration: dropping redundant tokens of the backbone while replenishing them in the feature pyramid without attention operations. 2) Token gathering: reducing computation cost caused by global attention by introducing few gathering tokens. The base/large-level Fast-iTPN achieve 88.75%/89.5% top-1 accuracy on ImageNet-1 K. With 1× training schedule using DINO, the base/large-level Fast-iTPN achieves 58.4%/58.8% box AP on COCO object detection, and a 57.5%/58.7% mIoU on ADE20 K semantic segmentation using MaskDINO. Fast-iTPN can accelerate the inference procedure by up to 70%, with negligible performance loss, demonstrating the potential to be a powerful backbone for downstream vision tasks.

In this paper, we present an integral pre-training framework based on masked image modeling (MIM). We advocate for pre-training the backbone and neck jointly so that the transfer gap between MIM and downstream recognition tasks is minimal. We make two technical contributions. First, we unify the reconstruction and recognition necks by inserting a feature pyramid into the pre-training stage. Second, we complement mask image modeling (MIM) with masked feature modeling (MFM) that offers multi-stage supervision to the feature pyramid. The pre-trained models, termed integrally pre-trained transformer pyramid networks (iTPNs), serve as powerful foundation models for visual recognition. In particular, the base/large-level iTPN achieves an 86.2%/87.8% top-1 accuracy on ImageNet-1K, a 53.2%/55.6% box AP on COCO object detection with 1× training schedule using Mask-RCNN, and a 54.7%/57.7% mIoU on ADE20K semantic segmentation using UPerHead – all these results set new records. Our work inspires the community to work on unifying upstream pre-training and downstream fine-tuning tasks. Code is available at github.com/sunsmarterjie/iTPN.

In 2D-to-3D pose estimation, it is important to exploit the spatial constraints of 2D joints, but it is not yet well modeled. To better model the relation of joints for 3D pose estimation, we propose an effective but simple net-work, called Graliormer11Codes:https://girhub.com/zhoawexi/GraFormer, where a novel transformer architecture is designed via embedding graph convolution layers after multi-head attention block. The proposed GraFormer is built by repeatedly stacking the GraAttention block and the ChebGConv block. The proposed GraAttention block is a new transformer block designed for processing graph-structured data, which is able to learn better features through capturing global information from all the nodes as well as the explicit adjacency structure of nodes. To model the implicit high-order connection relations among non-neighboring nodes, the ChebGConv block is introduced to exchange information between non-neighboring nodes and attain a larger receptive field. We have empirically shown the superiority of GraFormer through extensive experiments on popular public datasets. Specifically, GraFormer outperforms the state-of-the-art GraghSH [38] on the Hu-man3.6M dataset yet only contains 18% parameters of it.

Object skeleton detection requires the convolutional neural networks to recognize objects and their parts in the cluttered background, overcome the image definition degradation brought by the pooling layers, and predict the location of skeleton pixels in different scale granularity. Most existing object skeleton detection methods take great efforts into the designing of side-output networks for multiscale feature fusion. Despite the great progress achieved by them, there are still many problems that hinder the development of object skeleton detection, such as the manually designed network is labor-intensive and the network initialization depends on models pretrained on large-scale datasets. To alleviate these issues, we propose a genetic NAS method to automatically search on a newly designed architecture search space for adaptive multiscale feature fusion. Furthermore, we introduce a symmetric encoder-decoder search space based on reversing the VGG network, in which the decoder can reuse the ImageNet pretrained model of VGG. The searched networks improve the performance of the state-of-the-art methods on commonly used skeleton detection benchmarks, which proves the efficacy of our method.

Conventional networks for object skeleton detection are usually hand-crafted. Despite the effectiveness, hand-crafted network architectures lack the theoretical basis and require intensive prior knowledge to implement representation complementarity for objects/parts in different granularity. In this paper, we propose an adaptive linear span network (AdaLSN), driven by neural architecture search (NAS), to automatically configure and integrate scale-aware features for object skeleton detection. AdaLSN is formulated with the theory of linear span, which provides one of the earliest explanations for multi-scale deep feature fusion. AdaLSN is materialized by defining a mixed unit-pyramid search space, which goes beyond many existing search spaces using unit-level or pyramid-level features. Within the mixed space, we apply genetic architecture search to jointly optimize unit-level operations and pyramid-level connections for adaptive feature space expansion. AdaLSN substantiates its versatility by achieving significantly higher accuracy and latency trade-off compared with the state-of-the-arts. It also demonstrates general applicability to image-to-mask tasks such as edge detection and road extraction. Code is available at https://github.com/sunsmarterjie/SDL-Skeletongithub.com/sunsmarterjie/SDL-Skeleton.