Abstract:Museums serve as vital repositories of cultural heritage and historical artifacts spanning diverse epochs, civilizations, and regions, preserving well-documented collections. Data reveal key attributes such as age, origin, material, and cultural significance. Understanding museum exhibits from their images requires reasoning beyond visual features. In this work, we facilitate such reasoning by (a) collecting and curating a large-scale dataset of 65M images and 200M question-answer pairs in the standard museum catalog format for exhibits from all around the world; (b) training large vision-language models on the collected dataset; (c) benchmarking their ability on five visual question answering tasks. The complete dataset is labeled by museum experts, ensuring the quality as well as the practical significance of the labels. We train two VLMs from different categories: the BLIP model, with vision-language aligned embeddings, but lacking the expressive power of large language models, and the LLaVA model, a powerful instruction-tuned LLM enriched with vision-language reasoning capabilities. Through exhaustive experiments, we provide several insights on the complex and fine-grained understanding of museum exhibits. In particular, we show that some questions whose answers can often be derived directly from visual features are well answered by both types of models. On the other hand, questions that require the grounding of the visual features in repositories of human knowledge are better answered by the large vision-language models, thus demonstrating their superior capacity to perform the desired reasoning. Find our dataset, benchmarks, and source code at: https://github.com/insait-institute/Museum-65
Abstract:The source-free cross-domain few-shot learning (CD-FSL) task aims to transfer pretrained models to target domains utilizing minimal samples, eliminating the need for source domain data. Addressing this issue requires models to have robust generalization abilities and strong feature representation, aligning with the characteristics of large-scale pretrained models. However, large-scale models tend to lose representational ability in cross-domain scenarios due to limited sample diversity. \zlh{Given the abundant diversity provided by semantic modality, this paper leverages textual modality to enhance training sample diversity with CLP model}, meanwhile improving model transfer efficiency. Specifically, we propose the SeGD-VPT framework, which is divided into two phases. The first step aims to increase feature diversity by adding diversity prompts to each support sample, thereby generating varying input and enhancing sample diversity. Furthermore, we use diversity descriptions of classes to guide semantically meaningful learning of diversity prompts, proposing random combinations and selections of texts to increase textual diversity. Additionally, deep prompt tuning is introduced to enhance the model's transfer capability. After training of the first step, support samples with different diversity prompts are input into the CLIP backbone to generate enhanced features. After generation, the second phase trains classifiers using the generated features. Extensive experimental results across several benchmarks verify our method is comparable to SOTA source-utilized models and attain the best performance under the source-free CD-FSL setting.
Abstract:In this paper, we focus on the Ego-Exo Object Correspondence task, an emerging challenge in the field of computer vision that aims to map objects across ego-centric and exo-centric views. We introduce ObjectRelator, a novel method designed to tackle this task, featuring two new modules: Multimodal Condition Fusion (MCFuse) and SSL-based Cross-View Object Alignment (XObjAlign). MCFuse effectively fuses language and visual conditions to enhance target object localization, while XObjAlign enforces consistency in object representations across views through a self-supervised alignment strategy. Extensive experiments demonstrate the effectiveness of ObjectRelator, achieving state-of-the-art performance on Ego2Exo and Exo2Ego tasks with minimal additional parameters. This work provides a foundation for future research in comprehensive cross-view object relation understanding highlighting the potential of leveraging multimodal guidance and cross-view alignment. Codes and models will be released to advance further research in this direction.
Abstract:Recent advances in image super-resolution (SR) have significantly benefited from the incorporation of Transformer architectures. However, conventional techniques aimed at enlarging the self-attention window to capture broader contexts come with inherent drawbacks, especially the significantly increased computational demands. Moreover, the feature perception within a fixed-size window of existing models restricts the effective receptive fields and the intermediate feature diversity. This study demonstrates that a flexible integration of attention across diverse spatial extents can yield significant performance enhancements. In line with this insight, we introduce Multi-Range Attention Transformer (MAT) tailored for SR tasks. MAT leverages the computational advantages inherent in dilation operation, in conjunction with self-attention mechanism, to facilitate both multi-range attention (MA) and sparse multi-range attention (SMA), enabling efficient capture of both regional and sparse global features. Further coupled with local feature extraction, MAT adeptly capture dependencies across various spatial ranges, improving the diversity and efficacy of its feature representations. We also introduce the MSConvStar module, which augments the model's ability for multi-range representation learning. Comprehensive experiments show that our MAT exhibits superior performance to existing state-of-the-art SR models with remarkable efficiency (~3.3 faster than SRFormer-light).
Abstract:Leveraging multiple sensors is crucial for robust semantic perception in autonomous driving, as each sensor type has complementary strengths and weaknesses. However, existing sensor fusion methods often treat sensors uniformly across all conditions, leading to suboptimal performance. By contrast, we propose a novel, condition-aware multimodal fusion approach for robust semantic perception of driving scenes. Our method, CAFuser uses an RGB camera input to classify environmental conditions and generate a Condition Token that guides the fusion of multiple sensor modalities. We further newly introduce modality-specific feature adapters to align diverse sensor inputs into a shared latent space, enabling efficient integration with a single and shared pre-trained backbone. By dynamically adapting sensor fusion based on the actual condition, our model significantly improves robustness and accuracy, especially in adverse-condition scenarios. We set the new state of the art with CAFuser on the MUSES dataset with 59.7 PQ for multimodal panoptic segmentation and 78.2 mIoU for semantic segmentation, ranking first on the public benchmarks.
Abstract:Reconstructing 3D visuals from functional Magnetic Resonance Imaging (fMRI) data, introduced as Recon3DMind in our conference work, is of significant interest to both cognitive neuroscience and computer vision. To advance this task, we present the fMRI-3D dataset, which includes data from 15 participants and showcases a total of 4768 3D objects. The dataset comprises two components: fMRI-Shape, previously introduced and accessible at https://huggingface.co/datasets/Fudan-fMRI/fMRI-Shape, and fMRI-Objaverse, proposed in this paper and available at https://huggingface.co/datasets/Fudan-fMRI/fMRI-Objaverse. fMRI-Objaverse includes data from 5 subjects, 4 of whom are also part of the Core set in fMRI-Shape, with each subject viewing 3142 3D objects across 117 categories, all accompanied by text captions. This significantly enhances the diversity and potential applications of the dataset. Additionally, we propose MinD-3D, a novel framework designed to decode 3D visual information from fMRI signals. The framework first extracts and aggregates features from fMRI data using a neuro-fusion encoder, then employs a feature-bridge diffusion model to generate visual features, and finally reconstructs the 3D object using a generative transformer decoder. We establish new benchmarks by designing metrics at both semantic and structural levels to evaluate model performance. Furthermore, we assess our model's effectiveness in an Out-of-Distribution setting and analyze the attribution of the extracted features and the visual ROIs in fMRI signals. Our experiments demonstrate that MinD-3D not only reconstructs 3D objects with high semantic and spatial accuracy but also deepens our understanding of how human brain processes 3D visual information. Project page at: https://jianxgao.github.io/MinD-3D.
Abstract:Object detection, particularly open-vocabulary object detection, plays a crucial role in Earth sciences, such as environmental monitoring, natural disaster assessment, and land-use planning. However, existing open-vocabulary detectors, primarily trained on natural-world images, struggle to generalize to remote sensing images due to a significant data domain gap. Thus, this paper aims to advance the development of open-vocabulary object detection in remote sensing community. To achieve this, we first reformulate the task as Locate Anything on Earth (LAE) with the goal of detecting any novel concepts on Earth. We then developed the LAE-Label Engine which collects, auto-annotates, and unifies up to 10 remote sensing datasets creating the LAE-1M - the first large-scale remote sensing object detection dataset with broad category coverage. Using the LAE-1M, we further propose and train the novel LAE-DINO Model, the first open-vocabulary foundation object detector for the LAE task, featuring Dynamic Vocabulary Construction (DVC) and Visual-Guided Text Prompt Learning (VisGT) modules. DVC dynamically constructs vocabulary for each training batch, while VisGT maps visual features to semantic space, enhancing text features. We comprehensively conduct experiments on established remote sensing benchmark DIOR, DOTAv2.0, as well as our newly introduced 80-class LAE-80C benchmark. Results demonstrate the advantages of the LAE-1M dataset and the effectiveness of the LAE-DINO method.
Abstract:With the emergence of a single large model capable of successfully solving a multitude of tasks in NLP, there has been growing research interest in achieving similar goals in computer vision. On the one hand, most of these generic models, referred to as generalist vision models, aim at producing unified outputs serving different tasks. On the other hand, some existing models aim to combine different input types (aka data modalities), which are then processed by a single large model. Yet, this step of combination remains specialized, which falls short of serving the initial ambition. In this paper, we showcase that such specialization (during unification) is unnecessary, in the context of RGB-X video object tracking. Our single model tracker, termed XTrack, can remain blind to any modality X during inference time. Our tracker employs a mixture of modal experts comprising those dedicated to shared commonality and others capable of flexibly performing reasoning conditioned on input modality. Such a design ensures the unification of input modalities towards a common latent space, without weakening the modality-specific information representation. With this idea, our training process is extremely simple, integrating multi-label classification loss with a routing function, thereby effectively aligning and unifying all modalities together, even from only paired data. Thus, during inference, we can adopt any modality without relying on the inductive bias of the modal prior and achieve generalist performance. Without any bells and whistles, our generalist and blind tracker can achieve competitive performance compared to well-established modal-specific models on 5 benchmarks across 3 auxiliary modalities, covering commonly used depth, thermal, and event data.
Abstract:This paper addresses the challenge of cross-domain few-shot object detection (CD-FSOD), aiming to develop an accurate object detector for novel domains with minimal labeled examples. While transformer-based open-set detectors e.g., DE-ViT~\cite{zhang2023detect} have excelled in both open-vocabulary object detection and traditional few-shot object detection, detecting categories beyond those seen during training, we thus naturally raise two key questions: 1) can such open-set detection methods easily generalize to CD-FSOD? 2) If no, how to enhance the results of open-set methods when faced with significant domain gaps? To address the first question, we introduce several metrics to quantify domain variances and establish a new CD-FSOD benchmark with diverse domain metric values. Some State-Of-The-Art (SOTA) open-set object detection methods are evaluated on this benchmark, with evident performance degradation observed across out-of-domain datasets. This indicates the failure of adopting open-set detectors directly for CD-FSOD. Sequentially, to overcome the performance degradation issue and also to answer the second proposed question, we endeavor to enhance the vanilla DE-ViT. With several novel components including finetuning, a learnable prototype module, and a lightweight attention module, we present an improved Cross-Domain Vision Transformer for CD-FSOD (CD-ViTO). Experiments show that our CD-ViTO achieves impressive results on both out-of-domain and in-domain target datasets, establishing new SOTAs for both CD-FSOD and FSOD. All the datasets, codes, and models will be released to the community.
Abstract:A comprehensive understanding of videos is inseparable from describing the action with its contextual action-object interactions. However, many current video understanding tasks prioritize general action classification and overlook the actors and relationships that shape the nature of the action, resulting in a superficial understanding of the action. Motivated by this, we introduce Open-vocabulary Video Relation Extraction (OVRE), a novel task that views action understanding through the lens of action-centric relation triplets. OVRE focuses on pairwise relations that take part in the action and describes these relation triplets with natural languages. Moreover, we curate the Moments-OVRE dataset, which comprises 180K videos with action-centric relation triplets, sourced from a multi-label action classification dataset. With Moments-OVRE, we further propose a crossmodal mapping model to generate relation triplets as a sequence. Finally, we benchmark existing cross-modal generation models on the new task of OVRE.