UAV image object detection based on self-attention guidance and global feature fusion

Image and Vision Computing, Journal Year: 2024, Volume and Issue: 151, P. 105262 - 105262

Published: Sept. 10, 2024

Language: Английский

---

MSM-TDE: multi-scale semantics mining and tiny details enhancement network for retinal vessel segmentation

Complex & Intelligent Systems, Journal Year: 2025, Volume and Issue: 11(1)

Published: Jan. 1, 2025

Language: Английский

---

Direction-guided network for retinal vessel segmentation in OCTA images

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 103, P. 107455 - 107455

Published: Jan. 8, 2025

Language: Английский

---

SegDRoWS: Segmentation of diabetic retinopathy lesions by a whole-stage multi-scale feature fusion network

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 105, P. 107581 - 107581

Published: Jan. 30, 2025

Language: Английский

---

Multi-Scale Three-Path Network (MSTP-Net): A new architecture for retinal vessel segmentation

Measurement, Journal Year: 2025, Volume and Issue: unknown, P. 117100 - 117100

Published: March 1, 2025

Language: Английский

---

MSRP-TODNet: a multi-scale reinforced region wise analyser for tiny object detection

BMC Research Notes, Journal Year: 2025, Volume and Issue: 18(1)

Published: April 30, 2025

Detecting small, faraway objects in real-time surveillance is challenging due to limited pixel representation, affecting classifier performance. Deep Learning (DL) techniques generate feature maps enhance detection, but conventional methods suffer from high computational costs. To address this, we propose Multi-Scale Region-wise Pixel Analysis with GAN for Tiny Object Detection (MSRP-TODNet). The model trained and tested on VisDrone VID 2019 MS-COCO datasets. First, images undergo two-fold pre-processing using Improved Wiener Filter (IWF) artifact removal Adjusted Contrast Enhancement Method (ACEM) blurring correction. Multi-Agent Reinforcement (MARL) algorithm splits the pre-processed image into four regions, analyzing each maps. These are processed by Enhanced Feature Pyramid Network (EFPN), which merges them a single map. Finally, Generative Adversarial (GAN) detects bounding boxes. Experimental results DOTA dataset demonstrate that MSRP-TODNet outperforms existing state-of-the-art methods. Specifically, it achieves an mAP @0.5 of 84.2%, @0.5:0.95 54.1%, F1-Score 84.0%, surpassing improved TPH-YOLOv5, YOLOv7-Tiny, DRDet margins 1.7%-6.1% detection framework's effectiveness accurate, small object UAV aerial imagery.

Language: Английский

---

Gabor-modulated depth separable convolution for retinal vessel segmentation in fundus images

Computers in Biology and Medicine, Journal Year: 2025, Volume and Issue: 188, P. 109789 - 109789

Published: Feb. 12, 2025

Language: Английский

---

SFIT-Net: Spatial Reconstruction Feature Interaction Transformer Retinal Vessel Segmentation Algorithm

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 106, P. 107688 - 107688

Published: Feb. 20, 2025

Language: Английский

---

State-of-the-Art Deep Learning Methods for Microscopic Image Segmentation: Applications to Cells, Nuclei, and Tissues

Journal of Imaging, Journal Year: 2024, Volume and Issue: 10(12), P. 311 - 311

Published: Dec. 6, 2024

Microscopic image segmentation (MIS) is a fundamental task in medical imaging and biological research, essential for precise analysis of cellular structures tissues. Despite its importance, the process encounters significant challenges, including variability conditions, complex structures, artefacts (e.g., noise), which can compromise accuracy traditional methods. The emergence deep learning (DL) has catalyzed substantial advancements addressing these issues. This systematic literature review (SLR) provides comprehensive overview state-of-the-art DL methods developed over past six years microscopic images. We critically analyze key contributions, emphasizing how specifically tackle challenges cell, nucleus, tissue segmentation. Additionally, we evaluate datasets performance metrics employed studies. By synthesizing current identifying gaps existing approaches, this not only highlights transformative potential enhancing diagnostic research efficiency but also suggests directions future research. findings study have implications improving methodologies applications, ultimately fostering better patient outcomes advancing scientific understanding.

Language: Английский

---

Liver vessel MRI image segmentation based on dual-path diffusion model

Journal of Radiation Research and Applied Sciences, Journal Year: 2024, Volume and Issue: 17(3), P. 101025 - 101025

Published: Aug. 3, 2024

Accurate segmentation of liver blood vessels in magnetic resonance imaging (MRI) images is a challenging task due to the complex tree-like structure and anisotropic diffusion properties vessels. To solve this problem, we propose new Dual-Path Diffusion Model (DPDM) framework. The framework consists two collaborative paths: local feature learning path based on convolution operations global context modeling transform blocks. Local encodes rich shape priors preserve spatial details, while captures long-distance dependencies enhance representation. In decoding phase, boundary features from are fused with ordinary decoding, which further enhances sensitivity. addition, leverage multi-task scheme jointly optimize vascular prediction tasks an end-to-end manner. Experiments retrospective clinical dataset demonstrate that proposed DPDM achieves excellent performance vessel task. Compared state-of-the-art methods, our approach achieved 6.0% 7.3% improvement Dice coefficient IoU index, respectively. Our offers promising solution for automated precision medicine.

Language: Английский

---