NSTL国家科技图书文献中心

1. MLSA‑YOLO: a multi‑level feature fusion and scale‑adaptive framework for small object detection NSTL国家科技图书文献中心

Jiayu Peng | Kai Lv ... - 《The Journal of Supercomputing》 - 2025,81(4) - 528.1～528.24 - 共24页

摘要： combined it with the proposed multi-level Feature | feature fusion, hindering theaccurate representation of | . Furthermore,to address the challenges in feature fusion, we | small objects, certain feature* extractionparadigms* | multi-scale features. As a result, even high

关键词： YOLOv8 | Small object detection | Multi-level feature fusion | Scaleadaptive

2. Facial expression recognition method based on multi-level feature fusion of high-resolution images NSTL国家科技图书文献中心

Li Wan | Wenzhi Cheng - 《International Journal of Biometrics》 - 2025,17(1/2) - 57～72 - 共16页

摘要： recognition method based on multi-level* feature fusion of* | processing. Secondly, extract multi-level* features of* | facial images, and then fuse multi-level* features* | decoupled data of facial expressions based on feature | fusion results. Lastly, compare decoupled

关键词： Facial images | Expression recognition | High-resolution images | Multi-level feature fusion

3. A multi-level feature fusion artificial neural network for classification of acoustic emission signals NSTL国家科技图书文献中心

Jinliang Huang | Zhaolin Zhu ... - 《Annals of the New York Academy of Sciences》 - 2025,1544(Feb.) - 223～241 - 共19页

摘要： in each branch, utilizing multi-level feature | In this paper, we introduce FUSION-ANN, a | acoustic emission (AE) signal classification. FUSION-ANN | . These features are concatenated to form a new feature | branches of FUSION-ANN. The network per-forms AE signal

关键词： artificial neural network | classification of AE signals | gammatone cepstral coefficient | linear predictive coding | Mel-frequency cepstral coefficients | multi-level feature fusion

4. Few-shot fault diagnosis for machinery using multi-scale perception multi-level feature fusion image quadrant entropy NSTL国家科技图书文献中心

Zhenya Wang | Pan Liang ... - 《Advanced engineering informatics》 - 2025,63(Jan.) - 102972.1～102972.15 - 共15页

摘要： based on multi-scale perception multi-level feature | multi-scale perception multi-level feature strategy is | fusion image quadrant entropy (MPMFFIQE). The MPMFFIQE | feature maps at various levels, maximizing the | fusion image quadrant entropy technique is proposed to

关键词： Machinery | Few-shot fault diagnosis | Multi-scale perception multi-level feature fusion image quadrant entropy | Support vector machine

5. Oracle Character Recognition Based on Attention Enhancement and Multi-level Feature Fusion NSTL国家科技图书文献中心

Zhiwang Han | Nurbiya Yadikar ... - 《Pattern Recognition,Part XXXI》 - International Conference on Pattern Recognition - 2025, - 13～28 - 共16页

摘要： multi-level feature fusion by adaptively fusing the | Enhancement (CFE) and Adaptive Multi-level Classifier Fusion | Character Feature* Enhancement Network (CFE-Net). The model* | consists of two key components: Character Feature | , thereby significantly enhancing high-level* semantic*

关键词： Oracle bone characters | Character image recognition | Attention mechanism | Feature fusion

6. Multi-level feature fusion networks for smoke recognition in remote sensing imagery NSTL国家科技图书文献中心

Wang, Yupeng | Wang, Yongli ... - 《Neural Networks》 - 2025,184 - Article 107112～Article 107112 - 共13页

摘要： challenges, we propose the Multi-level Feature Fusion | Bilinear Feature* Fusion Module combines these enriched* | contrastive learning. MFFNet begins by extracting multi | in smoke appearance. The Attention Feature | Enhancement Module further refines these multi-scale

关键词： Remote sensing imagery | Smoke recognition | Feature fusion | Contrastive learning | Deep learning

7. Real-Time Text Detection with Multi-level Feature Fusion and Pixel Clustering NSTL国家科技图书文献中心

Lu Xu | Zhufeng Jiang ... - 《Pattern Recognition and Computer Vision,Part VII》 - Chinese Conference on Pattern Recognition and Computer Vision - 2025, - 16～29 - 共14页

摘要：, named the Multi-Level* Feature Fusion and Pixel* | context understanding and refines lower-level feature | extensibility of pixel-level* representations, but they face* | . This method utilizes a lightweight feature* extraction* | network, enhanced with specially designed Feature

关键词： Real-time text detection | Feature enhancement | Pixel clustering

8. MDFF-Net: Multi-level Dynamic Feature Fusion Network Combined with Deep Supervision Mechanism for Low-Light Image Enhancement NSTL国家科技图书文献中心

Xuxu Yang | Yitao Liang ... - 《Circuits, systems, and signal processing》 - 2025,44(2) - 1371～1399 - 共29页

摘要： Multi-level Dynamic Feature Fusion Module (MDFFM) to | images, we develop a multilevel dynamic feature fusion | specifically, we propose a multi-scale input encoder (MIE) in | the feature* extraction part, which aims to help the* | , and minimize redundant features during the fusion

关键词： Multi-level feature fusion | Dynamic fusion | Low-light image enhancement | Deep supervision

9. AMFT-YOLO: A Adaptive Multi-scale YOLO Algorithm with Multi-level Feature Fusion for Object Detection in UAV Scenes NSTL国家科技图书文献中心

Tiebiao Wang | Zhenchao Cui ... - 《MultiMedia Modeling,Part I》 - International Conference on MultiMedia Modeling - 2025, - 72～85 - 共14页

摘要：, termed Adaptive Multi-Scale Feature* Tower-YOLO (AMFT* | adaptive feature* fusion. The effectiveness of our method* | classification tasks on the feature* maps. This allows for the* | efficient utilization of limited feature* information* | . Secondly, we proposed the Bidirectional Multi-Scale Skip

关键词： Unmanned aerial vehicles | Missing information | Cluttered background | Multi-Scale

10. TSAM: Multi-level Condition-Enhanced Image Inpainting via Progressive Feature Fusion and Decoding Mechanism NSTL国家科技图书文献中心

Xiongfei Jia | Jiahao Meng ... - 《Sixteenth International Conference on Graphics and Image Processing (ICGIP 2024)》 - International Conference on Graphics and Image Processing - 2025, - 135391A.1～135391A.12 - 共12页

摘要： Multi-level Conditional Encoding (MCE) network in the | Progressive Feature* Fusion and Decoding (PFFD) network is* | In recent years, deep learning-based image | inpainting methods have made significant progress by | incorporating structural priors. Due to the lack of

摘要： combined it with the proposed multi-level Feature | feature fusion, hindering theaccurate representation of | . Furthermore,to address the challenges in feature fusion, we | small objects, certain feature* extractionparadigms* | multi-scale features. As a result, even high

关键词： YOLOv8 | Small object detection | Multi-level feature fusion | Scaleadaptive

摘要： recognition method based on multi-level* feature fusion of* | processing. Secondly, extract multi-level* features of* | facial images, and then fuse multi-level* features* | decoupled data of facial expressions based on feature | fusion results. Lastly, compare decoupled

关键词： Facial images | Expression recognition | High-resolution images | Multi-level feature fusion

摘要： in each branch, utilizing multi-level feature | In this paper, we introduce FUSION-ANN, a | acoustic emission (AE) signal classification. FUSION-ANN | . These features are concatenated to form a new feature | branches of FUSION-ANN. The network per-forms AE signal

关键词： artificial neural network | classification of AE signals | gammatone cepstral coefficient | linear predictive coding | Mel-frequency cepstral coefficients | multi-level feature fusion

摘要： based on multi-scale perception multi-level feature | multi-scale perception multi-level feature strategy is | fusion image quadrant entropy (MPMFFIQE). The MPMFFIQE | feature maps at various levels, maximizing the | fusion image quadrant entropy technique is proposed to

关键词： Machinery | Few-shot fault diagnosis | Multi-scale perception multi-level feature fusion image quadrant entropy | Support vector machine

摘要： multi-level feature fusion by adaptively fusing the | Enhancement (CFE) and Adaptive Multi-level Classifier Fusion | Character Feature* Enhancement Network (CFE-Net). The model* | consists of two key components: Character Feature | , thereby significantly enhancing high-level* semantic*

关键词： Oracle bone characters | Character image recognition | Attention mechanism | Feature fusion

摘要： challenges, we propose the Multi-level Feature Fusion | Bilinear Feature* Fusion Module combines these enriched* | contrastive learning. MFFNet begins by extracting multi | in smoke appearance. The Attention Feature | Enhancement Module further refines these multi-scale

关键词： Remote sensing imagery | Smoke recognition | Feature fusion | Contrastive learning | Deep learning

摘要：, named the Multi-Level* Feature Fusion and Pixel* | context understanding and refines lower-level feature | extensibility of pixel-level* representations, but they face* | . This method utilizes a lightweight feature* extraction* | network, enhanced with specially designed Feature

关键词： Real-time text detection | Feature enhancement | Pixel clustering

摘要： Multi-level Dynamic Feature Fusion Module (MDFFM) to | images, we develop a multilevel dynamic feature fusion | specifically, we propose a multi-scale input encoder (MIE) in | the feature* extraction part, which aims to help the* | , and minimize redundant features during the fusion

关键词： Multi-level feature fusion | Dynamic fusion | Low-light image enhancement | Deep supervision

关键词： Unmanned aerial vehicles | Missing information | Cluttered background | Multi-Scale

摘要： Multi-level Conditional Encoding (MCE) network in the | Progressive Feature* Fusion and Decoding (PFFD) network is* | In recent years, deep learning-based image | inpainting methods have made significant progress by | incorporating structural priors. Due to the lack of

关键词： Image inpainting | Twin-stream adversarial model | Multi-level conditional encoding | Progressive feature fusion and decoding

4008-161-200 800-990-8900

国家科技图书文献中心

摘要： combined it with the proposed multi-level Feature | feature fusion, hindering theaccurate representation of | . Furthermore,to address the challenges in feature fusion, we | small objects, certain feature extractionparadigms | multi-scale features. As a result, even high

关键词： YOLOv8 | Small object detection | Multi-level feature fusion | Scaleadaptive

摘要： recognition method based on multi-level feature fusion of | processing. Secondly, extract multi-level features of | facial images, and then fuse multi-level features | decoupled data of facial expressions based on feature | fusion results. Lastly, compare decoupled

关键词： Facial images | Expression recognition | High-resolution images | Multi-level feature fusion

摘要： in each branch, utilizing multi-level feature | In this paper, we introduce FUSION-ANN, a | acoustic emission (AE) signal classification. FUSION-ANN | . These features are concatenated to form a new feature | branches of FUSION-ANN. The network per-forms AE signal

关键词： artificial neural network | classification of AE signals | gammatone cepstral coefficient | linear predictive coding | Mel-frequency cepstral coefficients | multi-level feature fusion

摘要： based on multi-scale perception multi-level feature | multi-scale perception multi-level feature strategy is | fusion image quadrant entropy (MPMFFIQE). The MPMFFIQE | feature maps at various levels, maximizing the | fusion image quadrant entropy technique is proposed to

关键词： Machinery | Few-shot fault diagnosis | Multi-scale perception multi-level feature fusion image quadrant entropy | Support vector machine

摘要： multi-level feature fusion by adaptively fusing the | Enhancement (CFE) and Adaptive Multi-level Classifier Fusion | Character Feature Enhancement Network (CFE-Net). The model | consists of two key components: Character Feature | , thereby significantly enhancing high-level semantic

关键词： Oracle bone characters | Character image recognition | Attention mechanism | Feature fusion

摘要： challenges, we propose the Multi-level Feature Fusion | Bilinear Feature Fusion Module combines these enriched | contrastive learning. MFFNet begins by extracting multi | in smoke appearance. The Attention Feature | Enhancement Module further refines these multi-scale

关键词： Remote sensing imagery | Smoke recognition | Feature fusion | Contrastive learning | Deep learning

摘要：, named the Multi-Level Feature Fusion and Pixel | context understanding and refines lower-level feature | extensibility of pixel-level representations, but they face | . This method utilizes a lightweight feature extraction | network, enhanced with specially designed Feature

关键词： Real-time text detection | Feature enhancement | Pixel clustering

摘要： Multi-level Dynamic Feature Fusion Module (MDFFM) to | images, we develop a multilevel dynamic feature fusion | specifically, we propose a multi-scale input encoder (MIE) in | the feature extraction part, which aims to help the | , and minimize redundant features during the fusion

关键词： Multi-level feature fusion | Dynamic fusion | Low-light image enhancement | Deep supervision

关键词： Unmanned aerial vehicles | Missing information | Cluttered background | Multi-Scale

摘要： Multi-level Conditional Encoding (MCE) network in the | Progressive Feature Fusion and Decoding (PFFD) network is | In recent years, deep learning-based image | inpainting methods have made significant progress by | incorporating structural priors. Due to the lack of

关键词： Image inpainting | Twin-stream adversarial model | Multi-level conditional encoding | Progressive feature fusion and decoding

4008-161-200 800-990-8900

国家科技图书文献中心

摘要： combined it with the proposed multi-level Feature | feature fusion, hindering theaccurate representation of | . Furthermore,to address the challenges in feature fusion, we | small objects, certain feature* extractionparadigms* | multi-scale features. As a result, even high

摘要： recognition method based on multi-level* feature fusion of* | processing. Secondly, extract multi-level* features of* | facial images, and then fuse multi-level* features* | decoupled data of facial expressions based on feature | fusion results. Lastly, compare decoupled

摘要： multi-level feature fusion by adaptively fusing the | Enhancement (CFE) and Adaptive Multi-level Classifier Fusion | Character Feature* Enhancement Network (CFE-Net). The model* | consists of two key components: Character Feature | , thereby significantly enhancing high-level* semantic*

摘要： challenges, we propose the Multi-level Feature Fusion | Bilinear Feature* Fusion Module combines these enriched* | contrastive learning. MFFNet begins by extracting multi | in smoke appearance. The Attention Feature | Enhancement Module further refines these multi-scale

摘要：, named the Multi-Level* Feature Fusion and Pixel* | context understanding and refines lower-level feature | extensibility of pixel-level* representations, but they face* | . This method utilizes a lightweight feature* extraction* | network, enhanced with specially designed Feature

摘要： Multi-level Dynamic Feature Fusion Module (MDFFM) to | images, we develop a multilevel dynamic feature fusion | specifically, we propose a multi-scale input encoder (MIE) in | the feature* extraction part, which aims to help the* | , and minimize redundant features during the fusion

摘要： Multi-level Conditional Encoding (MCE) network in the | Progressive Feature* Fusion and Decoding (PFFD) network is* | In recent years, deep learning-based image | inpainting methods have made significant progress by | incorporating structural priors. Due to the lack of