NSTL国家科技图书文献中心

1. Window-Based Channel Attention for Wavelet-Enhanced Learned Image Compression NSTL国家科技图书文献中心

Heng Xu | Bowen Hai ... - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 450～467 - 共18页

摘要：Learned Image Compression (LIC) models have | achieved superior rate-distortion performance than | traditional codecs. Existing LIC models use CNN, Transformer | , or Mixed CNN-Transformer as basic blocks. However | , limited by the shifted window attention, Swin

关键词： Learned image compression | Window-based channel attention | Receptive field | Wavelet transform

2. COCA: Classifier-Oriented Calibration via Textual Prototype for Source-Free Universal Domain Adaptation NSTL国家科技图书文献中心

Xinghong Liu | Yi Zhou - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 337～353 - 共17页

摘要：Universal domain adaptation (UniDA) aims to | address domain and category shifts across data sources | . Recently, due to more stringent data restrictions | , researchers have introduced source-free UniDA (SF-UniDA). SF | -UniDA methods eliminate the need for direct access to

关键词： Source-Free universal domain adaptation | Transfer learning | Few-Shot learning

3. A Universal Structure of YOLO Series Small Object Detection Models NSTL国家科技图书文献中心

Shengchao Hu | Xiao Liu ... - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 468～484 - 共17页

摘要：The YOLO series detection models play a | crucial role in target detection tasks. However, these | models are typically trained on datasets with standard | angles. For datasets like Visdron2021 and Tinyperson | , there are challenges related to small, dense, and

关键词： YOLO series model | Small object detection | Universal structure

4. CNN Mixture-of-Depths NSTL国家科技图书文献中心

Rinor Cakaj | Jens Mehnert ... - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 148～166 - 共19页

摘要：We introduce Mixture-of-Depths (MoD) for | Convolutional Neural Networks (CNNs), a novel approach that | enhances the computational efficiency of CNNs by | selectively processing channels based on their relevance to | the current prediction. This method optimizes

关键词： CNN | Mixture-of-Depths | Computational efficiency | Inference speed

5. D'OH: Decoder-Only Random Hypernetworks for Implicit Neural Representations NSTL国家科技图书文献中心

Cameron Gordon | Lachlan E. MacDonald ... - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 128～147 - 共20页

摘要：Deep implicit functions have been found to be | an effective tool for efficiently encoding all | manner of natural signals. Their attractiveness stems | from their ability to compactly represent signals | with little to no offline training data. Instead

关键词： Implicit neural representations | Compression | Hypernetworks

6. FOTV-HQS: A Fractional-Order Total Variation Model for LiDAR Super-Resolution with Deep Unfolding Network NSTL国家科技图书文献中心

Huiying Xi | Xia Yuan ... - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 76～92 - 共17页

摘要：LiDAR super-resolution can improve the quality | of point cloud data, which is critical for | improving many downstream tasks such as object detection | , identification, and tracking. Traditional LiDAR super | -resolution models often struggle with issues like block

关键词： Super-resolution | LiDAR | Fractional-order total variation | Deep unfolding

7. EQ-CBM: A Probabilistic Concept Bottleneck with Energy-Based Models and Quantized Vectors NSTL国家科技图书文献中心

Sangwon Kim | Dasom Ahn ... - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 270～286 - 共17页

摘要：The demand for reliable AI systems has | intensified the need for interpretable deep neural networks | . Concept bottleneck models (CBMs) have gained attention | as an effective approach by leveraging human | -understandable concepts to enhance interpretability. However

关键词： Concept bottleneck model | Energy-based model | Vector quantization

8. SAMIF: Adapting Segment Anything Model for Image Inpainting Forensics NSTL国家科技图书文献中心

Lan Zhang | Xinshan Zhu ... - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 303～319 - 共17页

摘要：Image inpainting technologies pose increasing | threats to the security of image data through malicious | use. Therefore, image inpainting forensics is | crucial. The Segment Anything Model (SAM) is a powerful | universal image segmentation model for various downstream

关键词： Image inpainting forensics | Segment anything model | Cross-domain alignment fusion | Artifact feature generator

9. HT-SSPG: Hierarchical Transformers for Semantic Surface Point Generation in 3D Object Detection NSTL国家科技图书文献中心

Wenhao Kong | Xiaowei Zhang - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 20～37 - 共18页

摘要：Currently, the incomplete point cloud | structure in LiDAR point clouds has become the primary | challenge for improving detector performance. Point cloud | completion methods address this issue by adding more points | to regions of interest, however, due to imprecise

关键词： 3-D object detection | Semantic point generation | Hierarchical transformers

10. UNet--: Memory-Efficient and Feature-Enhanced Network Architecture Based on U-Net with Reduced Skip-Connections NSTL国家科技图书文献中心

Lingxiao Yin | Wei Tao ... - 《Computer Vision - ACCV 2024,Part VII》 - Asian Conference on Computer Vision - 2025, - 185～201 - 共17页

摘要：U-Net models with encoder, decoder, and skip | -connections components have demonstrated effectiveness in a | variety of vision tasks. The skip-connections transmit | fine-grained information from the encoder to the | decoder. It is necessary to maintain the feature maps

摘要：Learned Image Compression (LIC) models have | achieved superior rate-distortion performance than | traditional codecs. Existing LIC models use CNN, Transformer | , or Mixed CNN-Transformer as basic blocks. However | , limited by the shifted window attention, Swin

关键词： Learned image compression | Window-based channel attention | Receptive field | Wavelet transform

摘要：Universal domain adaptation (UniDA) aims to | address domain and category shifts across data sources | . Recently, due to more stringent data restrictions | , researchers have introduced source-free UniDA (SF-UniDA). SF | -UniDA methods eliminate the need for direct access to

关键词： Source-Free universal domain adaptation | Transfer learning | Few-Shot learning

摘要：The YOLO series detection models play a | crucial role in target detection tasks. However, these | models are typically trained on datasets with standard | angles. For datasets like Visdron2021 and Tinyperson | , there are challenges related to small, dense, and

关键词： YOLO series model | Small object detection | Universal structure

摘要：We introduce Mixture-of-Depths (MoD) for | Convolutional Neural Networks (CNNs), a novel approach that | enhances the computational efficiency of CNNs by | selectively processing channels based on their relevance to | the current prediction. This method optimizes

关键词： CNN | Mixture-of-Depths | Computational efficiency | Inference speed

摘要：Deep implicit functions have been found to be | an effective tool for efficiently encoding all | manner of natural signals. Their attractiveness stems | from their ability to compactly represent signals | with little to no offline training data. Instead

关键词： Implicit neural representations | Compression | Hypernetworks

摘要：LiDAR super-resolution can improve the quality | of point cloud data, which is critical for | improving many downstream tasks such as object detection | , identification, and tracking. Traditional LiDAR super | -resolution models often struggle with issues like block

关键词： Super-resolution | LiDAR | Fractional-order total variation | Deep unfolding

摘要：The demand for reliable AI systems has | intensified the need for interpretable deep neural networks | . Concept bottleneck models (CBMs) have gained attention | as an effective approach by leveraging human | -understandable concepts to enhance interpretability. However

关键词： Concept bottleneck model | Energy-based model | Vector quantization

摘要：Image inpainting technologies pose increasing | threats to the security of image data through malicious | use. Therefore, image inpainting forensics is | crucial. The Segment Anything Model (SAM) is a powerful | universal image segmentation model for various downstream

关键词： Image inpainting forensics | Segment anything model | Cross-domain alignment fusion | Artifact feature generator

摘要：Currently, the incomplete point cloud | structure in LiDAR point clouds has become the primary | challenge for improving detector performance. Point cloud | completion methods address this issue by adding more points | to regions of interest, however, due to imprecise

关键词： 3-D object detection | Semantic point generation | Hierarchical transformers

摘要：U-Net models with encoder, decoder, and skip | -connections components have demonstrated effectiveness in a | variety of vision tasks. The skip-connections transmit | fine-grained information from the encoder to the | decoder. It is necessary to maintain the feature maps

关键词： Skip-connection | Memory | U-Net

4008-161-200 800-990-8900

国家科技图书文献中心