Remote sensing foundation models largely break away from the traditional paradigm of designing task-specific models, offering greater scalability across multiple tasks. However, they face challenges such as low computational efficiency and limited interpretability, especially when dealing with large-scale remote sensing images. To overcome these, we draw inspiration from heat conduction, a physical process modeling local heat diffusion. Building on this idea, we are the first to explore the potential of using the parallel computing model of heat conduction to simulate the local region correlations in high-resolution remote sensing images, and introduce RS-vHeat, an efficient multi-modal remote sensing foundation model. Specifically, RS-vHeat 1) applies the Heat Conduction Operator (HCO) with a complexity of O(N^ 1.5 ) and a global receptive field, reducing computational overhead while capturing remote sensing object structure information to guide heat diffusion; 2) learns the frequency distribution representations of various scenes through a self-supervised strategy based on frequency domain hierarchical masking and multi-domain reconstruction; 3) significantly improves efficiency and performance over state-of-the-art techniques across 4 tasks and 10 datasets. Compared to attention-based remote sensing foundation models, we reduce memory usage by 84%, FLOPs by 24% and improves throughput by 2.7 times. The code will be made publicly available.
论文检索
输入标题、作者或关键词，从 2,578 篇学术成果中精准定位
In this work, we introduce JDCL - a new method for continual learning with generative rehearsal based on joint diffusion models. Neural networks suffer from catastrophic forgetting defined as abrupt loss in the model's performance when retrained with additional data coming from a different distribution. Generative-replay-based continual learning methods try to mitigate this issue by retraining a model with a combination of new and rehearsal data sampled from a generative model. In this work, we propose to extend this idea by combining a continually trained classifier with a diffusion-based generative model into a single - jointly optimized neural network. We show that such shared parametrization, combined with the knowledge distillation technique allows for stable adaptation to new tasks without catastrophic forgetting. We evaluate our approach on several benchmarks, where it outperforms recent state-of-the-art generative replay techniques. Additionally, we extend our method to the semi-supervised continual learning setup, where it outperforms competing buffer-based replay techniques, and evaluate, in a self-supervised manner, the quality of trained representations.
Unsupervised object discovery (UOD) aims to detect and segment objects in 2D images without handcrafted annotations. Recent progress in self-supervised representation learning has led to some success in UOD algorithms. However, the absence of ground truth provides existing UOD methods with two challenges: 1) determining if a discovered region is foreground or background, and 2) knowing how many objects remain undiscovered. To address these two problems, previous solutions rely on foreground priors to distinguish if the discovered region is foreground, and conduct one or fixed iterations of discovery. However, the existing foreground priors are heuristic and not always robust, and a fixed number of discoveries leads to under or over-segmentation, since the number of objects in images varies. This paper introduces UnionCut, a robust and well-grounded foreground prior based on min-cut and ensemble methods that detects the union of foreground areas of an image, allowing UOD algorithms to identify foreground objects and stop discovery once the majority of the foreground union in the image is segmented. In addition, we propose UnionSeg, a distilled transformer of UnionCut that outputs the foreground union more efficiently and accurately. Our experiments show that by combining with UnionCut or UnionSeg, previous state-of-the-art UOD methods witness an increase in the performance of single object discovery, saliency detection and self-supervised instance segmentation on various benchmarks. The code is available at https://github.com/YFaris/UnionCut.
Gaussian-based World Model: Gaussian Priors for Voxel-Based Occupancy Prediction and Future Motion Prediction
PDF ↗In autonomous driving, accurately predicting occupancy and motion is crucial for safe navigation within dynamic environments. However, existing methods often suffer from difficulties in handling complex scenes and uncertainty arising from sensor data. To address these issues, we propose a new Gaussian-based World Model (GWM), seamlessly integrating raw multi-modal sensor inputs. In 1st stage, Gaussian representation learner utilizes self-supervised pretraining to learn robust Gaussian representation. Gaussian representation integrates semantic and geometric information and establishes a robust probabilistic understanding of the environment. In 2nd stage, GWM seamlessly integrates learning, simulation, and planning into a unified framework, empowering the uncertainty-aware simulator & planner to jointly forecast future scene evolutions and vehicle trajectories. Simulator generates future scene predictions by modeling both static and dynamic elements, while planner calculates optimal paths to minimize collision risks, thus enhancing navigation safety. Overall, GWM employs a sensor-to-planning world model that directly processes raw sensor data, setting it apart from previous methods. Experiments show that GWM outperforms state-of-the-art approaches by 1.46% in semantic comprehension and 0.07m in motion prediction. Moreover, we provide an in-depth analysis of Gaussian representations under complex scenarios.
Light Field (LF) images captured under low illumination conditions typically exhibit low quality. Recent learning-based methods for low-light LF enhancement are generally tailored to specific illumination inputs, limiting their performance in real-world scenes. Moreover, how to maintain the inherent view-consistency in the enhanced images also remain as a difficult problem. In this paper, we propose to explore the view consistency for scene-adaptive low-light LF enhancement. We first analyze the view consistency for LF illumination maps and design a self-supervised view-consistent loss to keep the consistency between the illumination maps of different views in LFs. To enhance the model's perception of illumination, we combine both global and local information to estimate the illumination map, which is easily plugged into other models. Subsequently, we use the illumination maps to light up the low-light LF images and restore the corruption to produce the final enhanced image. Extensive experiments demonstrate that our View-Consistenct Network (VCNet) outperforms state-of-the-art methods on real-world low-light LF datasets in both fixed lighting conditions and dynamic lighting conditions. Our proposed illumination adjustment is also demonstrated that can comprehensively improve the performance of existing methods in terms of both image quality and view consistency.
Embodied Image Captioning: Self-supervised Learning Agents for Spatially Coherent Image Descriptions
PDF ↗We present a self-supervised method to improve an agent's abilities in describing arbitrary objects while actively exploring a generic environment. This is a challenging problem, as current models struggle to obtain coherent image captions due to different camera viewpoints and clutter. We propose a three-phase framework to fine-tune existing captioning models that enhances caption accuracy and consistency across views via a consensus mechanism. First, an agent explores the environment, collecting noisy image-caption pairs. Then, a consistent pseudo-caption for each object instance is distilled via consensus using a large language model. Finally, these pseudo-captions are used to fine-tune an off-the-shelf captioning model, with the addition of contrastive learning. We analyse the performance of the combination of captioning models, exploration policies, pseudo-labeling methods, and fine-tuning strategies, on our manually labeled test set. Results show that a policy can be trained to mine samples with higher disagreement compared to classical baselines. Our pseudo-captioning method, in combination with all policies, has a higher semantic similarity compared to other existing methods, and fine-tuning improves caption accuracy and consistency by a significant margin. Code and test set annotations available at https://hsp-iit.github.io/embodied-captioning.
ModalTune: Fine-Tuning Slide-Level Foundation Models with Multi-Modal Information for Multi-task Learning in Digital Pathology
PDF ↗Prediction tasks in digital pathology are challenging due to the massive size of whole-slide images (WSIs) and the weak nature of training signals. Advances in computing, data availability, and self-supervised learning (SSL) have paved the way for slide-level foundation models (SLFMs) that can improve prediction tasks in low-data regimes. However, current methods under-utilize shared information between tasks and modalities. To overcome this challenge, we propose ModalTune, a novel fine-tuning framework which introduces the Modal Adapter to integrate new modalities without modifying SLFM weights. Additionally, we use large-language models (LLMs) to encode labels as text, capturing semantic relationships across multiple tasks and cancer types in a single training recipe. ModalTune achieves state-of-the-art (SOTA) results against both uni-modal and multi-modal models across four cancer types, jointly improving survival and cancer subtype prediction while remaining competitive in pan-cancer settings. Additionally, we show ModalTune is generalizable to two out-of-distribution (OOD) datasets. To our knowledge, this is the first unified fine-tuning framework for multi-modal, multi-task, and pan-cancer modeling in digital pathology.
AI-generated face detectors trained via supervised learning typically rely on synthesized images from specific generators, limiting their generalization to emerging generative techniques. To overcome this limitation, we introduce a self-supervised method based on bi-level optimization. In the inner loop, we pretrain a vision encoder only on photographic face images using a set of linearly weighted pretext tasks: classification of categorical exchangeable image file format (EXIF) tags, ranking of ordinal EXIF tags, and detection of artificial face manipulations. The outer loop then optimizes the relative weights of these pretext tasks to enhance the coarse-grained detection of manipulated faces, serving as a proxy task for identifying AI-generated faces. In doing so, it aligns self-supervised learning more closely with the ultimate goal of AI-generated face detection. Once pretrained, the encoder remains fixed, and AI-generated faces are detected either as anomalies under a Gaussian mixture model fitted to photographic face features or by a lightweight two-layer perceptron serving as a binary classifier. Extensive experiments demonstrate that our detectors significantly outperform existing approaches in both one-class and binary classification settings, exhibiting strong generalization to unseen generators.
Absolute depth estimation from single camera sequence of images is a relevant task given that mobile machines increasingly rely on vision to navigate. Deep learning for stereo matching has been demonstrated to improve performance for stereo rectified depth estimation but these methods require straightforward left-right camera setups. This work proposes to introduce deep stereo matching to two views of a monocular image sequence obtained from a camera in motion in a static scene. This paper introduces a novel and principled spherical epipolar rectification model, which handles all camera motions. This rectification model is differentiable and allows self-supervised deep stereo matching algorithms to compute disparity and recover depth, given known camera pose. This paper also introduces a spherical crop operation which limits rectified image size and allows for competitive absolute depth estimation performance. This results in a spherical rectification model that is demonstrated to provide metric depth and compete favorably with current state-of-the-art monocular depth estimators. The code is available at https://gitlab.com/labv3d/spherical-stereo.git.
PAN-sharpening aims to fuse high-resolution panchromatic (PAN) images with low-resolution multi-spectral (MS) images to generate high-resolution multi-spectral (HRMS) outputs. However, cross-modality misalignment---caused by sensor placement, acquisition timing, and resolution disparity---induces a fundamental challenge. Conventional deep learning methods assume perfect pixel-wise alignment and rely on per-pixel reconstruction losses, leading to spectral distortion, double edges, and blurring when misalignment is present. To address this, we propose PAN-Crafter, a modality-consistent alignment framework that explicitly mitigates the misalignment gap between PAN and MS modalities. At its core, Modality-Adaptive Reconstruction (MARs) enables a single network to jointly reconstruct HRMS and PAN images, leveraging PAN's high-frequency details as auxiliary self-supervision. Additionally, we introduce Cross-Modality Alignment-Aware Attention (CM3A), a novel mechanism that bidirectionally aligns MS texture to PAN structure and vice versa, enabling adaptive feature refinement across modalities. Extensive experiments on multiple benchmark datasets demonstrate that our PAN-Crafter outperforms the most recent state-of-the-art method in all metrics, even with 50.11xfaster inference time and 0.63xthe memory size. Furthermore, it demonstrates strong generalization performance on unseen satellite datasets, showing its robustness across different conditions.
Traditional alignment methods for Large Vision and Language Models (LVLMs) primarily rely on human-curated preference data. Human-generated preference data is costly; machine-generated preference data is limited in quality; and self-supervised preference data often introduces hallucinations. To overcome these limitations, we propose a novel Panel-of-Peers learning framework inspired by collaborative learning among humans. This approach leverages a panel of LVLMs, each evaluating and learning from their collective outputs through an iterative self-improvement process. By simulating a peer review system, our models generate, assess, and refine outputs in response to a curated set of prompts, mimicking a classroom learning environment. We demonstrate that this methodology enhances model performance without requiring extensive human-labeled datasets. Our experiments show significant improvement across multiple benchmarks, demonstrating the potential of peer evaluations as a scalable alternative to self-supervised alignment. Notably, we show that Panel-of-Peers increases the average score on fifteen benchmarks from 48% to 57%.
Medical image retrieval is essential for clinical diagnosis and medical education, yet remains highly challenging in endoscopic imaging due to limited annotated data, the lack of domain-specific pretrained models, and subtle visual similarities across anatomical regions. In this work, we utilize self-supervised contrastive learning to pretrain a strong image encoder tailored for endoscopic data, which serves as the backbone for downstream retrieval tasks. For text-to-image retrieval, we adopt a multi-modal contrastive learning approach that aligns textual and visual representations based on this pretrained backbone. To further enhance retrieval performance, we propose a novel re-ranking module that leverages the reasoning capabilities of large vision-language models (LVLMs), such as GPT-4o and Gemini. We also provide a comparative analysis of various retrieval strategies, offering insights into their effectiveness in clinical scenarios. Our method achieves top-2 in text-to-image and top-5 in image-to-image retrieval at the ENTRep Challenge 2025, demonstrating its potential value for endoscopic image retrieval. Source code is available at https://github.com/ELO-Lab/ENTRep-LDSF.
The rapid development of audio-driven talking head generators and advanced Text-To-Speech (TTS) models has led to more sophisticated temporal deepfakes. These advances highlight the need for robust methods capable of detecting and localizing deepfakes, even under novel, unseen attack scenarios. Current state-of-the-art deepfake detectors, while accurate, are often computationally expensive and struggle to generalize to novel manipulation techniques. To address these challenges, we propose multimodal approaches for the AV-Deepfake1M 2025 challenge. For the visual modality, we leverage handcrafted features to improve interpretability and adaptability. For the audio modality, we adapt a self-supervised learning (SSL) backbone coupled with graph attention networks to capture rich audio representations, improving detection robustness. Our approach strikes a balance between performance and real-world deployment, focusing on resilience and potential interpretability. On the AV-Deepfake1M++ dataset, our multimodal system achieves AUC of 92.78% for deepfake classification task and IoU of 0.3536 for temporal localization using only the audio modality.
Advances in Generative AI have made video-level deepfake detection increasingly challenging, exposing the limitations of current detection techniques. In this paper, we present HOLA, our solution to the Video-Level Deepfake Detection track of 2025 1M-Deepfakes Detection Challenge. Inspired by the success of large-scale pre-training in the general domain, we first scale audio-visual self-supervised pre-training in the multimodal video-level deepfake detection, which leverages our self-built dataset of 1.81M samples, thereby leading to a unified two-stage framework. To be specific, HOLA features an iterative-aware cross-modal learning module for selective audio-visual interactions, hierarchical contextual modeling with gated aggregations under the local-global perspective, and a pyramid-like refiner for scale-aware cross-grained semantic enhancements. Moreover, we propose the pseudo supervised singal injection strategy to further boost model performance. Extensive experiments across expert models and MLLMs impressivly demonstrate the effectiveness of our proposed HOLA. We also conduct a series of ablation studies to explore the crucial design factors of our introduced components. Remarkably, our HOLA ranks 1st, outperforming the second by 0.0476 AUC on the TestA set.
Medical Foundation Models (MFMs) are revolutionizing radiography image analysis with scalable and generalized diagnostic capabilities. However, their effectiveness in real-world clinical practice is limited due to insufficient interpretability. To address this limitation, we propose RadLAS, a novel MFM for interpretable Radiographic image analysis by introducing Lesion-Aware Self-supervised pre-training. Unlike conventional MFMs that rely on post-hoc explanations, RadLAS innovates by directly emulating human diagnostic reasoning to first grounding lesion evidence and then making decisions accordingly. Specifically, RadLAS introduces two self-supervised tasks: (I) Lesion-grounded Reconstruction, which learns structured anatomical representations by restoring lesion-aware image patches into their healthy counterparts, thereby facilitating pixel-level grounding of lesion evidence via input-normal contrast. (II) Lesion-discrimination Contrastive Learning, which enhances lesion-aware pattern in representations by explicitly decoupling grounded lesion evidence as clinical cues and aligning them with global semantics, thereby enabling direct lesion-oriented diagnosis while preserving global context. RadLAS demonstrates excellent performance across diverse downstream radiographic datasets, offering verifiable explanations by deriving specific diagnoses (Task II) based on grounded lesion evidence (Task I), while preserving generalized representations essential for high diagnostic accuracy. Extensive experiments demonstrate that RadLAS (i) achieves superior interpretability with highly correlated lesion prediction and localization, surpassing 11 interpretable medical models; (ii) delivers scalable representation learning, outperforming 14 SOTA supervised and self-supervised MFMs.
Image deocclusion (or amodal completion) aims to recover the invisible regions (\ie, shape and appearance) of occluded instances in images. Despite recent advances, the scarcity of high-quality data that balances diversity, plausibility, and fidelity remains a major obstacle. To address this challenge, we identify three critical elements: leveraging in-the-wild image data for diversity, incorporating human expertise for plausibility, and utilizing generative priors for fidelity. We propose SynergyAmodal, a novel framework for co-synthesizing in-the-wild amodal datasets with comprehensive shape and appearance annotations, which integrates these elements through a tripartite data-human-model collaboration. First, we design an occlusion-grounded self-supervised learning algorithm to harness the diversity of in-the-wild image data, fine-tuning an inpainting diffusion model into a partial completion diffusion model. Second, we establish a co-synthesis pipeline to iteratively filter, refine, select, and annotate the initial deocclusion results of the partial completion diffusion model, ensuring plausibility and fidelity through human expert guidance and prior model constraints. This pipeline generates a high-quality paired amodal dataset with extensive category and scale diversity, comprising approximately 16K pairs. Finally, we train a full completion diffusion model on the synthesized dataset, incorporating text prompts as conditioning signals. Extensive experiments demonstrate the effectiveness of our framework in achieving zero-shot generalization and textual controllability. Our code, dataset, and models will be made publicly available at https://github.com/imlixinyang/SynergyAmodal.
Graph classification is a fundamental machine learning problem with extensive applications in multimedia and biochemical analysis. Contemporary graph classification models usually require precise graph labels for supervision, even after self-supervised pre-training. However, in practical applications, the extensive precise annotation of graphs could be expensive or impractical. To exploit data efficiently, this work studies partial label graph learning, in which each graph is linked to a set of candidate labels but only one of them is accurate. Label ambiguity would bring difficulties in extracting graph semantics and the risk of overfitting noisy partial labels. Here, we present a novel approach called Coupled Dual Separation (CODE). To improve graph semantics mining under label ambiguity, our CODE contains a message passing branch and a graph kernel branch, which explore graph semantics implicitly and explicitly, respectively. To facilitate information exchange, we utilize one branch to separate partially labeled graphs into an informative set and an uninformative set, which provides guidance for the optimization of the other branch. Furthermore, to mitigate the risk of overfitting, parameters in coupled branches are partitioned into critical and non-critical ones for separated optimization procedures. Extensive experiments on several benchmark datasets validate the effectiveness of the proposed CODE.
Text-to-Motion Retrieval (TMR) is a challenging task to retrieve relevant motion sequences with the natural language description. Existing TMR methods primarily utilize single embeddings to represent and align text and motion sequences. However, real-world motion sequences typically contain multiple sequential actions with intricate semantics, which are hard to precisely capture by single embedding. Additionally, relying solely on naive contrastive training to capture high-level semantics may struggle to perceive and capture fine-grained action details necessary for precise text-motion alignment. In this work, we propose a novel Sequence-Event Semantic Consistent Learning (SECL) framework for 3D human motion retrieval. Specifically, we introduce a self-supervised learning strategy to incorporate fine-grained action details into the motion representations via the generative feedback from the diffusion model. We design a parameter-free sequence-level interaction to explore coarse-grained alignment and an event-level interaction that utilizes several learnable queries to capture event semantics in a shared learning manner for fine-grained alignment. Furthermore, an inter-consistency loss is introduced to align the event semantics between the motion and corresponding text, and an intra-diversity loss is designed to encourage event features to attend to different contents, effectively capturing the rich action information. Finally, we modify the traditional contrastive alignment objective and propose an importance-sampling strategy to emphasize harder negatives for discriminative representation learning. Extensive experiments show that our method significantly outperforms existing methods in text-to-motion retrieval and other challenging tasks, e.g., human interaction recognition and motion temporal localization.
Electroencephalogram (EEG) signal classification faces significant challenges due to data distribution shifts caused by heterogeneous electrode configurations, acquisition protocols, and hardware discrepancies across domains. This paper introduces IMAC, a novel channel-dependent mask and imputation self-supervised framework that formulates the alignment of cross-domain EEG data shifts as a spatial time series imputation task. To address heterogeneous electrode configurations in cross-domain scenarios, IMAC first standardizes different electrode layouts using a 3D-to-2D positional unification mapping strategy, establishing unified spatial representations. Unlike previous mask-based self-supervised representation learning methods, IMAC introduces spatio-temporal signal alignment. This involves constructing a channel-dependent mask and reconstruction task framed as a low-to-high resolution EEG spatial imputation problem. Consequently, this approach simulates cross-domain variations such as channel omissions and temporal instabilities, thus enabling the model to leverage the proposed imputer for robust signal alignment during inference. Furthermore, IMAC incorporates a disentangled structure that separately models the temporal and spatial information of the EEG signals separately, reducing computational complexity while enhancing flexibility and adaptability. Comprehensive evaluations across 10 publicly available EEG datasets demonstrate IMAC's superior performance, achieving state-of-the-art classification accuracy in both cross-subject and cross-center validation scenarios. Notably, IMAC shows strong robustness under both simulated and real-world distribution shifts, surpassing baseline methods by up to 35% in integrity scores while maintaining consistent classification accuracy.
The electrocardiogram (ECG) is an essential and effective tool for diagnosing heart diseases. However, its effectiveness can be compromised by noise or unavailability of one or more leads of the standard 12-lead recordings, resulting in diagnostic errors or uncertainty. To address these challenges, we propose TolerantECG, a foundation model for ECG signals that is robust to noise and capable of functioning with arbitrary subsets of the standard 12-lead ECG. TolerantECG training combines contrastive and self-supervised learning frameworks to jointly learn ECG signal representations alongside their corresponding knowledge-retrieval-based text report descriptions and corrupted or lead-missing signals. Comprehensive benchmarking results demonstrate that TolerantECG consistently ranks as the best or second-best performer across various ECG signal conditions and class levels in the PTB-XL dataset, and achieves the highest performance on the MIT-BIH Arrhythmia Database. The source is available at this link: https://github.com/Fsoft-AIC/TolerantECG