DCPNet: A Densely Connected Pyramid Network for Monocular Depth Estimation

Lai, Zhitong; Tian, Rui; Wu, Zhiguo; Ding, Nannan; Sun, Linjian; Wang, Yanjie

doi:10.3390/s21206780

Cited by 4 publications

(3 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CNN-based UNet architecture has become more frequently employed in medical image processing, including classification and segmentation [16,17], in the encoder-decoder CNN design known as UNet, which is based on ignoring the encoder and decoder layers interconnection. These skip connections aim to learn abstract representations of the input data without losing the high-quality features of the input image [18]. Several studies have employed UNet and UNet-based algorithms for BT segmentation, with the BraTS dataset as their image source.…”

Section: Literature Reviewmentioning

confidence: 99%

High-Resolution Model for Segmenting and Predicting Brain Tumor Based on Deep UNet with Multi Attention Mechanism

2024

IJIES

View full text Add to dashboard Cite

The research presents a new technique for segmenting brain tumors using the UNet framework enhanced with an attention mechanism. By incorporating attention processes that selectively emphasize prominent aspects while recording comprehensive contextual information, our strategy overcomes the challenges of brain tumor delineation. The suggested UNet-attention model is intended to outperform traditional segmentation techniques regarding precision and clinical applicability. Integrating spatial and channel attention processes into the UNet design is one of our study's significant achievements. The spatial attention mechanism's focus improves the capacity of the model to differentiate the mechanism's tumor and non-tumor areas. Also, incorporating contextual clues from multi-scale hierarchies allows for a thorough comprehension of visual properties. The discrete wavelet transform has been applied as a feature extraction method to enhance the model performance regarding time and memory consumption. A wide range of datasets is evaluated in-depth, proving our UNet-attention model's superiority. Advanced deep learning is made possible by combining attention processes and contextual data to delineate tumors precisely and clinically. Many evaluation criteria involving dice scores, accuracy, mean IoU, sensitivity, specificity, and Hausdorff distance have been applied to evaluate our model performance in different aspects. The model attained a dice coefficient of 0.9971. The model's specificity of 0.9988 is particularly noteworthy, demonstrating its exceptional ability to identify regions without tumors accurately. The model also achieved 0.9986 accuracies, 0.9142 mean IoU, Hausdorff distance (mm) 3.48. These evaluation values were obtained for applying our model on flair images from BraTS 2020.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

High-Resolution Model for Segmenting and Predicting Brain Tumor Based on Deep UNet with Multi Attention Mechanism

2024

IJIES

View full text Add to dashboard Cite

show abstract

“…Neck: Adopts the CSP2 structure from CSPNet to bolster feature fusion capabilities. This structure, rooted in Feature Pyramid Network (FPN) [33,34] and Path Aggregation Network (PAN), performs multi-scale fusion by effectively integrating feature information from different scales [35].…”

Section: Yolov5 Networkmentioning

confidence: 99%

Fusion Enhancement of YOLOv5 and Copula Bayesian Classifier for Hand Gesture Recognition in Smart Sports Venues

Han,

Hao,

Yuan

et al. 2024

IEEE Access

View full text Add to dashboard Cite

This study aims addresses the challenge of gesture recognition in smart sports stadiums through the fusion enhancement of YOLOv5 and Copula Bayesian Classifier algorithm. By leveraging FasterNet for YOLOv5 transformation, integrating the Convolutional Block Attention Module (CBAM) as an attention mechanism, and merging it with the Copula Bayesian Classifier, we propose a tandem model for robust gesture recognition in the complex environments of sports venues. Extensive training on a substantial dataset yields an efficient and accurate gesture recognition model. Our findings showcase the effectiveness of the proposed algorithm, achieving a remarkable accuracy rate of 99.2% in identifying gesture categories swiftly. This advancement holds significant implications for future human-computer interaction in smart sports stadiums, offering a more seamless and intelligent mode of interaction for both spectators and athletes. Ultimately, it enhances user experience and management efficiency within smart sports venues. INDEX TERMSCopula Bayesian Classifier, Fusion enhancement of YOLOv5, Human-computer interaction, Hand gesture recognition, Image processing.

show abstract

“…With this method, the decoding process can be broken down into different components in order to maximize the benefits of good coding features. In [31], Lai, et al proposed a densely connected pyramid network for monocular depth estimation. By using dense connection modules, they not only integrate the features of adjacent floors, but also integrate the features of non-adjacent floors.…”

Section: The Network Combining Fpn and U-netmentioning

confidence: 99%

Depth Estimation Using Feature Pyramid U-Net and Polarized Self-Attention for Road Scenes

et al. 2022

View full text Add to dashboard Cite

Studies have shown that the observed image texture details and semantic information are of great significance for the depth estimation on the road scenes. However, there are ambiguities and inaccuracies in the boundary information of observed objects in previous methods. For this reason, we hope to design a new depth estimation method that can obtain higher accuracy and more accurate boundary information of the detected object. Based on polarized self-attention (PSA) and feature pyramid U-net, we proposed a new self-supervised monocular depth estimation model to extract more accurate texture details and semantic information. Firstly, we add a PSA module at the end of the depth encoder and pose encoder so that the network can extract more accurate semantic information. Then, based on the U-net, we put the multi-scale image obtained by the object detection module FPN (Feature Pyramid network) directly into the decoder. It can guide the model to learn semantic information, thus enhancing the boundary of the image. We evaluated our method on KITTI 2015 datasets and Make3D datasets, and our model achieved better results than previous studies. In order to verify the generalization of the model, we have done monocular, stereo, monocular plus stereo experiments. The experimental results show that our model has achieved better results in several main evaluation indexes and clearer boundary information. In order to compare different forms of PSA mechanism, we did ablation experiments. Compared with no PSA module, after adding the PSA module, better results in evaluating indicators were achieved. We also found that our model is better in monocular training than stereo training and monocular plus stereo training.

show abstract

DCPNet: A Densely Connected Pyramid Network for Monocular Depth Estimation

Cited by 4 publications

References 61 publications

High-Resolution Model for Segmenting and Predicting Brain Tumor Based on Deep UNet with Multi Attention Mechanism

High-Resolution Model for Segmenting and Predicting Brain Tumor Based on Deep UNet with Multi Attention Mechanism

Fusion Enhancement of YOLOv5 and Copula Bayesian Classifier for Hand Gesture Recognition in Smart Sports Venues

Depth Estimation Using Feature Pyramid U-Net and Polarized Self-Attention for Road Scenes

Contact Info

Product

Resources

About