NSTL国家科技图书文献中心

1. Open-Vocabulary Camouflaged Object Segmentation NSTL国家科技图书文献中心

Youwei Pang | Xiaoqi Zhao ... - 《Computer Vision - ECCV 2024,Part XLVII》 - European Conference on Computer Vision - 2025, - 476～495 - 共20页

摘要： for the challenging open-vocabulary* dense prediction* | introduce a new task, open-vocabulary* camouflaged object* | previous state-of-the-arts of open-vocabulary* semantic* | further expand the research on open-vocabulary* dense* | the way towards open-world object perception. Many

关键词： Open vocabulary | Camouflaged object segmentation

2. Diffusion Models for Open-Vocabulary Segmentation NSTL国家科技图书文献中心

Laurynas Karazija | Iro Laina ... - 《Computer Vision - ECCV 2024,Part V》 - European Conference on Computer Vision - 2025, - 299～317 - 共19页

摘要：Open-vocabulary segmentation is the task of | significant advances in open-vocabulary* segmentation, but at* | , making them applicable in an open-vocabulary* setting* | diffusion models for unsupervised open-vocabulary | segmenting anything that can be named in an image. Recently

关键词： Open-vocabulary segmentation | Vision-language

3. Open Vocabulary Multi-label Video Classification NSTL国家科技图书文献中心

Rohit Gupta | Mamshad Nayeem Rizve ... - 《Computer Vision - ECCV 2024,Part XXXIX》 - European Conference on Computer Vision - 2025, - 276～293 - 共18页

摘要： enabled significant progress in open* vocabulary computer* | focused on extending VLMs to open* vocabulary single* | the video in an open* vocabulary setting. We* | formulate this problem as open* vocabulary multi-label* | open vocabulary performance with two key

关键词： Open vocabulary | Multi-Modal | Video understanding

4. Anytime Continual Learning for Open Vocabulary Classification NSTL国家科技图书文献中心

Zhen Zhu | Yiming Gong ... - 《Computer Vision - ECCV 2024,Part VI》 - European Conference on Computer Vision - 2025, - 269～285 - 共17页

摘要： learning (AnytimeCL) for open* vocabulary image* | predictions of a partially fine-tuned model and a fixed open | vocabulary model that enables continual improvement when | We propose an approach for anytime continual | classification. The AnytimeCL problem aims to break away from

关键词： Continual learning | Anytime learning | Open-vocabulary classification

5. Open-Vocabulary RGB-Thermal Semantic Segmentation NSTL国家科技图书文献中心

Guoqiang Zhao | Junjie Huang ... - 《Computer Vision - ECCV 2024,Part LXXIV》 - European Conference on Computer Vision - 2025, - 304～320 - 共17页

摘要：. First, they do not have the open-vocabulary | open-vocabulary segmentation capability. OpenRSS | the open-vocabulary* capability by jointly utilizing* | effective open-vocabulary* semantic segmentation ability on* | RGB open-vocabulary* semantic segmentation methods on*

关键词： Open-vocabulary | RGB-T semantic segmentation | Prompt learning

6. Open-Vocabulary Instance Segmentation-Boundary IS-Goal NSTL国家科技图书文献中心

Quan Tang - 《Pattern Recognition and Computer Vision,Part IV》 - Chinese Conference on Pattern Recognition and Computer Vision - 2025, - 421～436 - 共16页

摘要： boundary detection and open-vocabulary* capabilities for* | open-vocabulary instance segmentation tasks on the MS | instance segmentation and boundary detection under open | -vocabulary. It includes a prompt text encoder, a two-stream | and boundary detection performance under open

关键词： Instance segmentation | Boundary detection | Open-vocabulary | Multi-modal fusion

7. SLAck: Semantic, Location, and Appearance Aware Open-Vocabulary Tracking NSTL国家科技图书文献中心

Siyuan Li | Lei Ke ... - 《Computer Vision - ECCV 2024,Part XXVII》 - European Conference on Computer Vision - 2025, - 1～18 - 共18页

摘要：Open-vocabulary Multiple Object Tracking (MOT | significantly for large-scale open-vocabulary* tracking* | the open-vocabulary* MOT and TAO TETA benchmarks. Our* | large-vocabulary* scenarios and unstable classification* | ) aims to generalize trackers to novel categories not

关键词： Open-Vocabulary | Multiple object tracking

8. O_2V-Mapping: Online Open-Vocabulary Mapping with Neural Implicit Representation NSTL国家科技图书文献中心

Muer Tie | Julong Wei ... - 《Computer Vision - ECCV 2024,Part LXXXVII》 - European Conference on Computer Vision - 2025, - 318～333 - 共16页

摘要：. However, implementing open-vocabulary* scene* | create an open-vocabulary* field, thus allowing for* | . Extensive experiments on open-vocabulary* object* | Online construction of open-ended language | scenes is crucial for robotic applications, where open

关键词： NeRF | Open-Vocabulary | Language scene construction

9. Open-Vocabulary Scene Graph Generation via Synonym-Based Predicate Descriptor NSTL国家科技图书文献中心

Yuta Goto | Satoshi Yamazaki ... - 《MultiMedia Modeling,Part III》 - International Conference on MultiMedia Modeling - 2025, - 142～156 - 共15页

摘要：This paper proposes a class balance-aware open | -vocabulary* scene graph generation (SGG) method that* | predicate inference. Existing SGG works address open | -vocabulary* relation predictions utilizing vision language* | -training and the open* world; there are many unseen*

关键词： Open-vocabulary scene graph generation

10. Collaborative Vision-Text Representation Optimizing for Open-Vocabulary Segmentation NSTL国家科技图书文献中心

Siyu Jiao | Hongguang Zhu ... - 《Computer Vision - ECCV 2024,Part XXXIII》 - European Conference on Computer Vision - 2025, - 399～416 - 共18页

摘要： challenging Open-Vocabulary* Segmentation (OVS) task* | popular OVS benchmarks. In open-vocabulary* semantic* | Pre-trained vision-language models, e.g. CLIP | , have been increasingly used to address the | , benefiting from their well-aligned vision-text embedding

摘要： for the challenging open-vocabulary* dense prediction* | introduce a new task, open-vocabulary* camouflaged object* | previous state-of-the-arts of open-vocabulary* semantic* | further expand the research on open-vocabulary* dense* | the way towards open-world object perception. Many

关键词： Open vocabulary | Camouflaged object segmentation

摘要：Open-vocabulary segmentation is the task of | significant advances in open-vocabulary* segmentation, but at* | , making them applicable in an open-vocabulary* setting* | diffusion models for unsupervised open-vocabulary | segmenting anything that can be named in an image. Recently

关键词： Open-vocabulary segmentation | Vision-language

摘要： enabled significant progress in open* vocabulary computer* | focused on extending VLMs to open* vocabulary single* | the video in an open* vocabulary setting. We* | formulate this problem as open* vocabulary multi-label* | open vocabulary performance with two key

关键词： Open vocabulary | Multi-Modal | Video understanding

摘要： learning (AnytimeCL) for open* vocabulary image* | predictions of a partially fine-tuned model and a fixed open | vocabulary model that enables continual improvement when | We propose an approach for anytime continual | classification. The AnytimeCL problem aims to break away from

关键词： Continual learning | Anytime learning | Open-vocabulary classification

摘要：. First, they do not have the open-vocabulary | open-vocabulary segmentation capability. OpenRSS | the open-vocabulary* capability by jointly utilizing* | effective open-vocabulary* semantic segmentation ability on* | RGB open-vocabulary* semantic segmentation methods on*

关键词： Open-vocabulary | RGB-T semantic segmentation | Prompt learning

摘要： boundary detection and open-vocabulary* capabilities for* | open-vocabulary instance segmentation tasks on the MS | instance segmentation and boundary detection under open | -vocabulary. It includes a prompt text encoder, a two-stream | and boundary detection performance under open

关键词： Instance segmentation | Boundary detection | Open-vocabulary | Multi-modal fusion

摘要：Open-vocabulary Multiple Object Tracking (MOT | significantly for large-scale open-vocabulary* tracking* | the open-vocabulary* MOT and TAO TETA benchmarks. Our* | large-vocabulary* scenarios and unstable classification* | ) aims to generalize trackers to novel categories not

关键词： Open-Vocabulary | Multiple object tracking

摘要：. However, implementing open-vocabulary* scene* | create an open-vocabulary* field, thus allowing for* | . Extensive experiments on open-vocabulary* object* | Online construction of open-ended language | scenes is crucial for robotic applications, where open

关键词： NeRF | Open-Vocabulary | Language scene construction

摘要：This paper proposes a class balance-aware open | -vocabulary* scene graph generation (SGG) method that* | predicate inference. Existing SGG works address open | -vocabulary* relation predictions utilizing vision language* | -training and the open* world; there are many unseen*

关键词： Open-vocabulary scene graph generation

摘要： challenging Open-Vocabulary* Segmentation (OVS) task* | popular OVS benchmarks. In open-vocabulary* semantic* | Pre-trained vision-language models, e.g. CLIP | , have been increasingly used to address the | , benefiting from their well-aligned vision-text embedding

关键词： Open-Vocabulary segmentation | Fine-tuning

4008-161-200 800-990-8900

国家科技图书文献中心

摘要： for the challenging open-vocabulary dense prediction | introduce a new task, open-vocabulary camouflaged object | previous state-of-the-arts of open-vocabulary semantic | further expand the research on open-vocabulary dense | the way towards open-world object perception. Many

关键词： Open vocabulary | Camouflaged object segmentation

摘要：Open-vocabulary segmentation is the task of | significant advances in open-vocabulary segmentation, but at | , making them applicable in an open-vocabulary setting | diffusion models for unsupervised open-vocabulary | segmenting anything that can be named in an image. Recently

关键词： Open-vocabulary segmentation | Vision-language

摘要： enabled significant progress in open vocabulary computer | focused on extending VLMs to open vocabulary single | the video in an open vocabulary setting. We | formulate this problem as open vocabulary multi-label | open vocabulary performance with two key

关键词： Open vocabulary | Multi-Modal | Video understanding

摘要： learning (AnytimeCL) for open vocabulary image | predictions of a partially fine-tuned model and a fixed open | vocabulary model that enables continual improvement when | We propose an approach for anytime continual | classification. The AnytimeCL problem aims to break away from

关键词： Continual learning | Anytime learning | Open-vocabulary classification

摘要：. First, they do not have the open-vocabulary | open-vocabulary segmentation capability. OpenRSS | the open-vocabulary capability by jointly utilizing | effective open-vocabulary semantic segmentation ability on | RGB open-vocabulary semantic segmentation methods on

关键词： Open-vocabulary | RGB-T semantic segmentation | Prompt learning

摘要： boundary detection and open-vocabulary capabilities for | open-vocabulary instance segmentation tasks on the MS | instance segmentation and boundary detection under open | -vocabulary. It includes a prompt text encoder, a two-stream | and boundary detection performance under open

关键词： Instance segmentation | Boundary detection | Open-vocabulary | Multi-modal fusion

摘要：Open-vocabulary Multiple Object Tracking (MOT | significantly for large-scale open-vocabulary tracking | the open-vocabulary MOT and TAO TETA benchmarks. Our | large-vocabulary scenarios and unstable classification | ) aims to generalize trackers to novel categories not

关键词： Open-Vocabulary | Multiple object tracking

摘要：. However, implementing open-vocabulary scene | create an open-vocabulary field, thus allowing for | . Extensive experiments on open-vocabulary object | Online construction of open-ended language | scenes is crucial for robotic applications, where open

关键词： NeRF | Open-Vocabulary | Language scene construction

摘要：This paper proposes a class balance-aware open | -vocabulary scene graph generation (SGG) method that | predicate inference. Existing SGG works address open | -vocabulary relation predictions utilizing vision language | -training and the open world; there are many unseen

关键词： Open-vocabulary scene graph generation

摘要： challenging Open-Vocabulary Segmentation (OVS) task | popular OVS benchmarks. In open-vocabulary semantic | Pre-trained vision-language models, e.g. CLIP | , have been increasingly used to address the | , benefiting from their well-aligned vision-text embedding

关键词： Open-Vocabulary segmentation | Fine-tuning

4008-161-200 800-990-8900

国家科技图书文献中心

摘要： for the challenging open-vocabulary* dense prediction* | introduce a new task, open-vocabulary* camouflaged object* | previous state-of-the-arts of open-vocabulary* semantic* | further expand the research on open-vocabulary* dense* | the way towards open-world object perception. Many

摘要：Open-vocabulary segmentation is the task of | significant advances in open-vocabulary* segmentation, but at* | , making them applicable in an open-vocabulary* setting* | diffusion models for unsupervised open-vocabulary | segmenting anything that can be named in an image. Recently

摘要： enabled significant progress in open* vocabulary computer* | focused on extending VLMs to open* vocabulary single* | the video in an open* vocabulary setting. We* | formulate this problem as open* vocabulary multi-label* | open vocabulary performance with two key

摘要： learning (AnytimeCL) for open* vocabulary image* | predictions of a partially fine-tuned model and a fixed open | vocabulary model that enables continual improvement when | We propose an approach for anytime continual | classification. The AnytimeCL problem aims to break away from

摘要：. First, they do not have the open-vocabulary | open-vocabulary segmentation capability. OpenRSS | the open-vocabulary* capability by jointly utilizing* | effective open-vocabulary* semantic segmentation ability on* | RGB open-vocabulary* semantic segmentation methods on*

摘要： boundary detection and open-vocabulary* capabilities for* | open-vocabulary instance segmentation tasks on the MS | instance segmentation and boundary detection under open | -vocabulary. It includes a prompt text encoder, a two-stream | and boundary detection performance under open

摘要：Open-vocabulary Multiple Object Tracking (MOT | significantly for large-scale open-vocabulary* tracking* | the open-vocabulary* MOT and TAO TETA benchmarks. Our* | large-vocabulary* scenarios and unstable classification* | ) aims to generalize trackers to novel categories not

摘要：. However, implementing open-vocabulary* scene* | create an open-vocabulary* field, thus allowing for* | . Extensive experiments on open-vocabulary* object* | Online construction of open-ended language | scenes is crucial for robotic applications, where open

摘要：This paper proposes a class balance-aware open | -vocabulary* scene graph generation (SGG) method that* | predicate inference. Existing SGG works address open | -vocabulary* relation predictions utilizing vision language* | -training and the open* world; there are many unseen*

摘要： challenging Open-Vocabulary* Segmentation (OVS) task* | popular OVS benchmarks. In open-vocabulary* semantic* | Pre-trained vision-language models, e.g. CLIP | , have been increasingly used to address the | , benefiting from their well-aligned vision-text embedding