NSTL国家科技图书文献中心

1. Contribution-Based Low-Rank Adaptation with Pre-training Model for Real Image Restoration NSTL国家科技图书文献中心

Dongwon Park | Hayeon Kim ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 87～105 - 共19页

摘要：Recently, pre-trained model and efficient | parameter tuning have achieved remarkable success in | natural language processing and high-level computer | vision with the aid of masked modeling and prompt | tuning. In low-level computer vision, however, there

关键词： Efficient fine-tuing | Low-rank adaptation | Pre-training

2. Depth-Guided NeRF Training via Earth Mover's Distance NSTL国家科技图书文献中心

Anita Rau | Josiah Aklilu ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 1～17 - 共17页

摘要：Neural Radiance Fields (NeRFs) are trained to | minimize the rendering loss of predicted viewpoints | . However, the photometric loss often does not provide | enough information to disambiguate between different | possible geometries yielding the same image. Previous

关键词： Neural radiance fields | Depth prediction | Monocular depth priors | Earth mover's distance

3. MMEarth: Exploring Multi-modal Pretext Tasks for Geospatial Representation Learning NSTL国家科技图书文献中心

Vishal Nedungadi | Ankit Kariryaa ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 164～182 - 共19页

摘要：The volume of unlabelled Earth observation (EO | ) data is huge, but many important applications lack | labelled training data. However, EO data offers the | unique opportunity to pair data from different | modalities and sensors automatically based on geographic

关键词： Representation learning | Self-supervised learning | Multi-modal | Multi-task | Masked autoencoder | Earth observation | Remote sensing | Satellite images | Sentinel-2

4. Elucidating the Hierarchical Nature of Behavior with Masked Autoencoders NSTL国家科技图书文献中心

Lucas Stoffl | Andy Bonnetto ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 106～125 - 共20页

摘要：Natural behavior is hierarchical. Yet, there | is a paucity of benchmarks addressing this aspect | . Recognizing the scarcity of large-scale hierarchical | behavioral benchmarks, we create a novel synthetic | basketball playing benchmark (Shot7M2). Beyond synthetic

关键词： Masked autoencoder | Action segmentation | Behavioral benchmarks | Synthetic movement data | Hierarchical representation learning | Behavioral hierarchy

5. SAH-SCI: Self-supervised Adapter for Efficient Hyperspectral Snapshot Compressive Imaging NSTL国家科技图书文献中心

Haijin Zeng | Yuxi Liu ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 311～328 - 共18页

摘要：Hyperspectral image (HSI) reconstruction is | vital for recovering spatial-spectral information from | compressed measurements in coded aperture snapshot spectral | imaging (CASSI) systems. Despite the effectiveness of | end-to-end and deep unfolding methods, their

关键词： Compressive imaging | Self-supervised learning | Adapter

6. LITA: Language Instructed Temporal-Localization Assistant NSTL国家科技图书文献中心

De-An Huang | Shijia Liao ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 202～218 - 共17页

摘要：There has been tremendous progress in | multimodal Large Language Models (LLMs). Recent works have | extended these models to video input with promising | instruction following capabilities. However, an important | missing piece is temporal localization. These models

关键词： Large language models | Video understanding | Temporal localization

7. Evolving Interpretable Visual Classifiers with Large Language Models NSTL国家科技图书文献中心

Mia Chiquier | Utkarsh Mall ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 183～201 - 共19页

摘要：Multimodal pre-trained models, such as CLIP | , are popular for zero-shot classification due to | their open-vocabulary flexibility and high performance | . However, vision-language models, which compute | similarity scores between images and class labels, are

关键词： Visual recognition | Interpretable representations

8. INTRA: Interaction Relationship-A ware Weakly Supervised Affordance Grounding NSTL国家科技图书文献中心

Ji Ha Jang | Hoigi Seo ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 18～34 - 共17页

摘要：Affordance denotes the potential interactions | inherent in objects. The perception of affordance can | enable intelligent agents to navigate and interact with | new environments efficiently. Weakly supervised | affordance grounding teaches agents the concept of

关键词： Affordance grounding | Weak supervision | Exocentric image | Contrastive learning | Interaction relation

9. Diagnosing and Re-learning for Balanced Multimodal Learning NSTL国家科技图书文献中心

Yake Wei | Siwei Li ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 71～86 - 共16页

摘要：To overcome the imbalanced multimodal learning | problem, where models prefer the training of specific | modalities, existing methods propose to control the | training of uni-modal encoders from different | perspectives, taking the inter-modal performance discrepancy

关键词： Multimodal learning | Learning state diagnosing | Re-learning

10. DEPICT: Diffusion-Enabled Permutation Importance for Image Classification Tasks NSTL国家科技图书文献中心

Sarah Jabbour | Gregory Kondas ... - 《Computer Vision - ECCV 2024,Part LXIV》 - European Conference on Computer Vision - 2025, - 35～51 - 共17页

摘要：We propose a permutation-based explanation | method for image classifiers. Current image-model | explanations like activation maps are limited to instance | -based explanations in the pixel space, making it | difficult to understand global model behavior. In contrast

摘要：Recently, pre-trained model and efficient | parameter tuning have achieved remarkable success in | natural language processing and high-level computer | vision with the aid of masked modeling and prompt | tuning. In low-level computer vision, however, there

关键词： Efficient fine-tuing | Low-rank adaptation | Pre-training

摘要：Neural Radiance Fields (NeRFs) are trained to | minimize the rendering loss of predicted viewpoints | . However, the photometric loss often does not provide | enough information to disambiguate between different | possible geometries yielding the same image. Previous

关键词： Neural radiance fields | Depth prediction | Monocular depth priors | Earth mover's distance

摘要：The volume of unlabelled Earth observation (EO | ) data is huge, but many important applications lack | labelled training data. However, EO data offers the | unique opportunity to pair data from different | modalities and sensors automatically based on geographic

关键词： Representation learning | Self-supervised learning | Multi-modal | Multi-task | Masked autoencoder | Earth observation | Remote sensing | Satellite images | Sentinel-2

摘要：Natural behavior is hierarchical. Yet, there | is a paucity of benchmarks addressing this aspect | . Recognizing the scarcity of large-scale hierarchical | behavioral benchmarks, we create a novel synthetic | basketball playing benchmark (Shot7M2). Beyond synthetic

关键词： Masked autoencoder | Action segmentation | Behavioral benchmarks | Synthetic movement data | Hierarchical representation learning | Behavioral hierarchy

摘要：Hyperspectral image (HSI) reconstruction is | vital for recovering spatial-spectral information from | compressed measurements in coded aperture snapshot spectral | imaging (CASSI) systems. Despite the effectiveness of | end-to-end and deep unfolding methods, their

关键词： Compressive imaging | Self-supervised learning | Adapter

摘要：There has been tremendous progress in | multimodal Large Language Models (LLMs). Recent works have | extended these models to video input with promising | instruction following capabilities. However, an important | missing piece is temporal localization. These models

关键词： Large language models | Video understanding | Temporal localization

摘要：Multimodal pre-trained models, such as CLIP | , are popular for zero-shot classification due to | their open-vocabulary flexibility and high performance | . However, vision-language models, which compute | similarity scores between images and class labels, are

关键词： Visual recognition | Interpretable representations

摘要：Affordance denotes the potential interactions | inherent in objects. The perception of affordance can | enable intelligent agents to navigate and interact with | new environments efficiently. Weakly supervised | affordance grounding teaches agents the concept of

关键词： Affordance grounding | Weak supervision | Exocentric image | Contrastive learning | Interaction relation

摘要：To overcome the imbalanced multimodal learning | problem, where models prefer the training of specific | modalities, existing methods propose to control the | training of uni-modal encoders from different | perspectives, taking the inter-modal performance discrepancy

关键词： Multimodal learning | Learning state diagnosing | Re-learning

摘要：We propose a permutation-based explanation | method for image classifiers. Current image-model | explanations like activation maps are limited to instance | -based explanations in the pixel space, making it | difficult to understand global model behavior. In contrast

关键词： Permutation importance | Explainable AI | Diffusion models

4008-161-200 800-990-8900

国家科技图书文献中心