Stacked U-Nets: A No-Frills Approach to Natural Image Segmentation

Shah, Sohil; Ghosh, Pallabi; Davis, Larry S.; Goldstein, Tom

doi:10.48550/arxiv.1804.10343

Cited by 8 publications

(8 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, such network is likely to be trapped into a sub-optimal solution because a stacked U-net is more complicated than a single U-net. As such, a stacked U-net is usually employed when there is a pre-train model [5] [6]or fed with a large number training data (more than 10K) [7]. There is a work concatenating two U-net by two loss [19], but they didn't consider the information sharing in two U-net.In this paper, we propose a bridging architecture between two U-nets.…”

Section: Network For Image Segmentationmentioning

confidence: 99%

Prostate Segmentation using 2D Bridged U-net

Chen

Zhang

et al. 2019

2019 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

In this paper, we focus on three problems in deep learning based medical image segmentation. Firstly, U-net, as a popular model for medical image segmentation, is difficult to train when convolutional layers increase even though a deeper network usually has a better generalization ability because of more learnable parameters. Secondly, the exponential ReLU (ELU), as an alternative of ReLU, is not much different from ReLU when the network of interest gets deep. Thirdly, the Dice loss, as one of the pervasive loss functions for medical image segmentation, is not effective when the prediction is close to ground truth and will cause oscillation during training. To address the aforementioned three problems, we propose and validate a deeper network that can fit medical image datasets that are usually small in the sample size. Meanwhile, we propose a new loss function to accelerate the learning process and a combination of different activation functions to improve the network performance. Our experimental results suggest that our network is comparable or superior to state-of-the-art methods.

show abstract

Section: Network For Image Segmentationmentioning

confidence: 99%

Prostate Segmentation using 2D Bridged U-net

Chen

Zhang

et al. 2019

2019 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

show abstract

“…The features of the building segmentation network are further used for super-resolution. For building segmentation we use the stacked U-Net (SUNET) architecture as proposed is [23] and adapted to satellite imagery in [24]. SUNET consists of 4 blocks of stacked U-Nets containing 2,7,7 and 1 U-Net module(s) respectively.…”

Section: Related Workmentioning

confidence: 99%

Unsupervised Super-Resolution of Satellite Imagery for High Fidelity Material Label Transfer

Ghosh

Ehrlich

Davis

et al. 2019

IGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium

View full text Add to dashboard Cite

Urban material recognition in remote sensing imagery is a highly relevant, yet extremely challenging problem due to the difficulty of obtaining human annotations, especially on low resolution satellite images. To this end, we propose an unsupervised domain adaptation based approach using adversarial learning. We aim to harvest information from smaller quantities of high resolution data (source domain) and utilize the same to super-resolve low resolution imagery (target domain). This can potentially aid in semantic as well as material label transfer from a richly annotated source to a target domain.

show abstract

“…Skipped connections between the contracting and expanding path help to preserve features from the input image. A variety of applications in the imaging field have harnessed the original or a modification of the U-Net structure [16][17][18][19][20][21][22]. Despite the advantages of these DL approaches in image reconstruction, there are still some limitations.…”

Section: Introductionmentioning

confidence: 99%

Two-step training deep learning framework for computational imaging without physics priors

Shang¹,

Hoffer-Hawlik²,

Luke

2021

Opt. Express

View full text Add to dashboard Cite

Deep learning (DL) is a powerful tool in computational imaging for many applications. A common strategy is to reconstruct a preliminary image as the input of a neural network to achieve an optimized image. Usually, the preliminary image is acquired with the prior knowledge of the model. One outstanding challenge, however, is that the model is sometimes difficult to acquire with high accuracy. Another main challenge is that the process to reconstruct the preliminary image is time consuming, which cannot achieve real-time imaging. In this work, a two-step-training DL (TST-DL) framework is proposed for real-time computational imaging without prior knowledge of the model. A single fully-connected layer (FCL) is trained to directly learn the model with the raw measurement data as input and the image as output. Then, this pre-trained FCL is fixed and connected with an un-trained deep convolutional network for a second-step training to improve the output image fidelity. This approach has three main advantages. First, no prior knowledge of the model is required since the first-step training is to directly learn the model. Second, real-time imaging can be achieved since the raw measurement data is directly used as the input to the model. Third, it can handle any dimension of the network input and solve the input-output dimension mismatch issues which arise in convolutional neural networks. We demonstrate this framework in the applications of single-pixel imaging and photoacoustic imaging for linear model cases. The results are quantitatively compared with those from other DL frameworks and model-based optimization approaches. Noise robustness and the required size of the training dataset are studied for this framework. We further extend this concept to nonlinear models in the application of image de-autocorrelation by using multiple FCLs in the first-step training. Overall, this TST-DL framework is widely applicable to many computational imaging techniques for real-time image reconstruction without the physics priors.

show abstract

Stacked U-Nets: A No-Frills Approach to Natural Image Segmentation

Cited by 8 publications

References 49 publications

Prostate Segmentation using 2D Bridged U-net

Prostate Segmentation using 2D Bridged U-net

Unsupervised Super-Resolution of Satellite Imagery for High Fidelity Material Label Transfer

Two-step training deep learning framework for computational imaging without physics priors

Contact Info

Product

Resources

About