Polyp-PVT: Polyp Segmentation with Pyramid Vision Transformers

CAAI Artificial Intelligence Research, Journal Year: 2023, Volume and Issue: unknown, P. 9150015 - 9150015

Published: June 30, 2023

Most polyp segmentation methods use convolutional neural networks (CNNs) as their backbone, leading to two key issues when exchanging information between the encoder and decoder: (1) taking into account differences in contribution different-level features, (2) designing an effective mechanism for fusing these features. Unlike existing CNN-based methods, we adopt a transformer encoder, which learns more powerful robust representations. In addition, considering image acquisition influence elusive properties of polyps, introduce three standard modules, including cascaded fusion module (CFM), camouflage identification (CIM), similarity aggregation (SAM). Among these, CFM is used collect semantic location polyps from high-level features; CIM applied capture disguised low-level SAM extends pixel features area with position entire area, thereby effectively cross-level The proposed model, named Polyp-PVT, suppresses noises significantly improves expressive capabilities. Extensive experiments on five widely adopted datasets show that model various challenging situations (e.g., appearance changes, small objects, rotation) than representative methods. available at https://github.com/DengPingFan/Polyp-PVT.

Language: Английский

---

Polyp-PVT: Polyp Segmentation with Pyramid Vision Transformers

arXiv (Cornell University), Journal Year: 2021, Volume and Issue: unknown

Published: Jan. 1, 2021

Most polyp segmentation methods use CNNs as their backbone, leading to two key issues when exchanging information between the encoder and decoder: 1) taking into account differences in contribution different-level features 2) designing an effective mechanism for fusing these features. Unlike existing CNN-based methods, we adopt a transformer encoder, which learns more powerful robust representations. In addition, considering image acquisition influence elusive properties of polyps, introduce three standard modules, including cascaded fusion module (CFM), camouflage identification (CIM), similarity aggregation (SAM). Among these, CFM is used collect semantic location polyps from high-level features; CIM applied capture disguised low-level features, SAM extends pixel area with position entire area, thereby effectively cross-level The proposed model, named Polyp-PVT, suppresses noises significantly improves expressive capabilities. Extensive experiments on five widely adopted datasets show that model various challenging situations (e.g., appearance changes, small objects, rotation) than representative methods. available at https://github.com/DengPingFan/Polyp-PVT.

Language: Английский

---

Polyp-SAM: transfer SAM for polyp segmentation

Medical Imaging 2018: Computer-Aided Diagnosis, Journal Year: 2024, Volume and Issue: unknown

Published: April 2, 2024

Automatic segmentation of colon polyps can significantly reduce the misdiagnosis cancer and improve physician annotation efficiency. While many methods have been proposed for polyp segmentation, training large-scale networks with limited colonoscopy data remains a challenge. Recently, Segment Anything Model (SAM) has recently gained much attention in both natural image medical segmentation. SAM demonstrates superior performance several vision benchmarks shows great potential In this study, we propose Poly-SAM, finetuned model compare its to state-of-the-art models. We also two transfer learning strategies without finetuning encoders. Evaluated on five public datasets, our Polyp-SAM achieves datasets impressive three dice scores all above 88%. This study adapting tasks.

Language: Английский

---

TCNet: Multiscale Fusion of Transformer and CNN for Semantic Segmentation of Remote Sensing Images

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2024, Volume and Issue: 17, P. 3123 - 3136

Published: Jan. 1, 2024

Semantic segmentation of remote sensing images plays a critical role in areas such as urban change detection, environmental protection, and geohazard identification. Convolutional Neural Networks (CNN) have been excessively employed for semantic over the past few years; however, limitation CNN is that there exists challenge extracting global context images, which vital segmentation, due to locality convolution operation. It informed recently developed Transformer equipped with powerful modeling capabilities. A network called TCNet proposed this study, parallel-in-branch architecture adopted TCNet. As such, takes advantage both CNN, low-level spatial details could be captured much shallower manner. In addition, novel fusion technique Interactive Self-attention (ISa) advanced fuse multi-level features extracted from branches. To bridge gap between regions, skip connection module Windowed Gating (WSaG) further added progressive upsampling network. Experiments on three public datasets (i.e., Bijie Landslide Dataset, WHU Building Massachusetts Buildings Dataset) depict yields superior performance state-of-the-art models. The IoU values obtained by these are 75.34% (ranked first among ten models compared), 91.16% thirteen 76.21% respectively.

Language: Английский

---

Uncertainty-Aware Hierarchical Aggregation Network for Medical Image Segmentation

IEEE Transactions on Circuits and Systems for Video Technology, Journal Year: 2024, Volume and Issue: 34(8), P. 7440 - 7453

Published: Feb. 26, 2024

Medical image segmentation is an essential process to assist clinics with computer-aided diagnosis and treatment. Recently, a large amount of convolutional neural network (CNN)-based methods have been rapidly developed achieved remarkable performances in several different medical tasks. However, the same type infected region or lesions often has diversity scales, making it challenging task achieve accurate segmentation. In this paper, we present novel Uncertainty-aware Hierarchical Aggregation Network, namely UHA-Net, for segmentation, which can fully make utilization cross-level multi-scale features handle scale variations. Specifically, propose hierarchical feature fusion (HFF) module aggregate high-level features, used produce global map coarse localization segmented target. Then, uncertainty-induced (UCF) fuse from adjacent levels, learn knowledge guidance capture contextual information resolutions. Further, aggregation (SAM) presented by using convolution kernels, effectively deal At last, formulate unified framework simultaneously inter-layer discriminability representations intra-layer leading results. We carry out experiments on three tasks, results demonstrate that our UHA-Net outperforms state-of-the-art methods. Our implementation code maps will be publicly at https://github.com/taozh2017/UHANet.

Language: Английский

---

Boundary Refinement Network for Colorectal Polyp Segmentation in Colonoscopy Images

IEEE Signal Processing Letters, Journal Year: 2024, Volume and Issue: 31, P. 954 - 958

Published: Jan. 1, 2024

Precise polyp segmentation is vitally essential for detection and diagnosis of early colorectal cancer. Recent advances in artificial intelligence have brought infinite possibilities this task. However, polyps usually vary greatly shape size contain ambiguous boundary, bringing tough challenges to precise segmentation. In letter, we introduce a novel Boundary Refinement Network (BRNet) To be specific, first boundary generation module (BGM) generate map by fusing both low-level spatial details high-level concepts. Then, utilize the boundary-guided refinement refine polyp-aware features at each layer with help cues from BGM prediction adjacent high layer. Through top-down deep supervision, our BRNet can localize regions accurately clear boundary. Extensive experiments are carried out on five datasets, results indicate effectiveness over seven recently reported methods.

Language: Английский

---

Meta-Polyp: A Baseline for Efficient Polyp Segmentation

Published: June 1, 2023

In recent years, polyp segmentation has gained significant importance, and many methods have been developed using CNN, Vision Transformer, Transformer techniques to achieve competitive results. However, these often face difficulties when dealing with out-of-distribution datasets, missing boundaries, small polyps. 2022, Meta-Former was introduced as a new baseline for vision, which not only improved the performance of multi-task computer vision but also addressed limitations CNN family backbones. To further enhance segmentation, we propose fusion UNet, along introduction Multi-scale Upsampling block level-up combination in decoder stage texture, Convformer base on idea Meta-former crucial information local feature. These blocks enable global information, such overall shape polyp, boundary is decision medical segmentation. Our proposed approach achieved obtained top result State Art CVC-300 dataset, Kvasir, CVC-ColonDB dataset. Apart from Kvasir-SEG, others are datasets.

Language: Английский

---

You only train twice: A lighter and faster method for industrial weld defect detection based on dynamic kernel network

Measurement, Journal Year: 2024, Volume and Issue: 231, P. 114642 - 114642

Published: April 3, 2024

Language: Английский

---

Towards Diverse Binary Segmentation via a Simple yet General Gated Network

International Journal of Computer Vision, Journal Year: 2024, Volume and Issue: 132(10), P. 4157 - 4234

Published: May 7, 2024

Language: Английский

---

ArSDM: Colonoscopy Images Synthesis with Adaptive Refinement Semantic Diffusion Models

Lecture notes in computer science, Journal Year: 2023, Volume and Issue: unknown, P. 339 - 349

Published: Jan. 1, 2023

Language: Английский

---