scholarly journals Comparing U-Net Based Models for Denoising Color Images

AI ◽  
2020 ◽  
Vol 1 (4) ◽  
pp. 465-487
Author(s):  
Rina Komatsu ◽  
Tad Gonsalves
Keyword(s):  
Deep Learning ◽  
Performance Improvement ◽  
Gaussian Noise ◽  
Salt And Pepper Noise ◽  
L1 Norm ◽  
Specific Kind ◽  
Wide Interface ◽  
Salt And Pepper ◽  
Deep Learning Model ◽  
And Training

Digital images often become corrupted by undesirable noise during the process of acquisition, compression, storage, and transmission. Although the kinds of digital noise are varied, current denoising studies focus on denoising only a single and specific kind of noise using a devoted deep-learning model. Lack of generalization is a major limitation of these models. They cannot be extended to filter image noises other than those for which they are designed. This study deals with the design and training of a generalized deep learning denoising model that can remove five different kinds of noise from any digital image: Gaussian noise, salt-and-pepper noise, clipped whites, clipped blacks, and camera shake. The denoising model is constructed on the standard segmentation U-Net architecture and has three variants—U-Net with Group Normalization, Residual U-Net, and Dense U-Net. The combination of adversarial and L1 norm loss function re-produces sharply denoised images and show performance improvement over the standard U-Net, Denoising Convolutional Neural Network (DnCNN), and Wide Interface Network (WIN5RB) denoising models.

scholarly journals Dropout, a basic and effective regularization method for a deep learning model: a case study

2021 ◽  
Vol 24 (2) ◽  
pp. 1009
Author(s):  
Brahim Jabir ◽  
Noureddine Falih
Keyword(s):  
Deep Learning ◽  
Weed Species ◽  
It Impacts ◽  
Artificial Neurons ◽  
Pest Detection ◽  
Learning Set ◽  
Validation Set ◽  
Deep Learning Model ◽  
And Training

Deep learning is based on a network of artificial neurons inspired by the human brain. This network is made up of tens or even hundreds of "layers" of neurons. The fields of application of deep learning are indeed multiple; Agriculture is one of those fields in which deep learning is used in various agricultural problems (disease detection, pest detection, and weed identification). A major problem with deep learning is how to create a model that works well, not only on the learning set but also on the validation set. Many approaches used in neural networks are explicitly designed to reduce overfit, possibly at the expense of increasing validation accuracy and training accuracy. In this paper, a basic technique (dropout) is proposed to minimize overfit, we integrated it into a convolutional neural network model to classify weed species and see how it impacts performance, a complementary solution (exponential linear units) are proposed to optimize the obtained results. The results showed that these proposed solutions are practical and highly accurate, enabling us to adopt them in deep learning models.

scholarly journals Light Actuation Based On Facial Mood Recognition

2020 ◽  
Vol 9 (05) ◽  
pp. 25052-25056
Author(s):  
Abhi Kadam ◽  
Anupama Mhatre ◽  
Sayali Redasani ◽  
Amit Nerurkar
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Convolutional Neural Network ◽  
Well Being ◽  
Training Data ◽  
Facial Emotion ◽  
Specific Data ◽  
Affective Meaning ◽  
Deep Learning Model ◽  
And Training

Current lighting technologies extend the options for changing the appearance of rooms and closed spaces, as such creating ambiences with an affective meaning. Using intelligence, these ambiences may instantly be adapted to the needs of the room’s occupant(s), possibly improving their well-being. In this paper, we set actuate lighting in our surrounding using mood detection. We analyze the mood of the person by Facial Emotion Recognition using deep learning model such as Convolutional Neural Network (CNN). On recognizing this emotion, we will actuate lighting in our surrounding in accordance with the mood. Based on implementation results, the system needs to be developed further by adding more specific data class and training data.

Adaptive Four-dot Median Filter for Removing 1-99% Densities of Salt-and-Pepper Noise in Images

2018 ◽  
Vol 11 (3) ◽  
pp. 47-61 ◽  
Author(s):  
Xin-Ming Zhang ◽  
Qiang Kang ◽  
Jin-Feng Cheng ◽  
Xia Wang
Keyword(s):  
Image Restoration ◽  
Performance Improvement ◽  
Adaptive Filtering ◽  
Median Filter ◽  
Window Size ◽  
Salt And Pepper Noise ◽  
Noise Detection ◽  
Median Filters ◽  
Simple Logic ◽  
Salt And Pepper

In order to accelerate denoising and improve the denoising performance of the current median filters, an Adaptive Four-dot Median Filter (AFMF) for image restoration is proposed in this article. AFMF is not only very efficient and fast in logic execution, but also it can restore the corrupted images with 1–99% densities of salt-and-pepper noise to the satisfactory ones. Without any complicated operation for noise detection, it intuitively and simply distinguishes impulse noises, while keeping the noise-free pixels intact. Only the uncorrupted pixels of the four-dot mask in adaptive filtering windows are used for the adoption of candidates for median finding, whatever filtering window size is. Furthermore, the adoption of recursive median filters leads to denoising performance improvement and faster filtering. The simple logic of the proposed algorithm obtains significant milestones on the fidelity of a restored image. Relevant experimental results on subjective visualization and objective digital measure validate the robustness of the proposed filter.

Feature Extraction from Noisy Image Using Intersecting Cortical Model

2010 ◽  
Vol 40-41 ◽  
pp. 516-522 ◽  
Author(s):  
Xiao Fei Wang ◽  
Bo Nian Li ◽  
Yan Li Huang ◽  
Xin Ran Wang
Keyword(s):  
Feature Extraction ◽  
Gaussian Noise ◽  
Feature Vector ◽  
Input Image ◽  
Noisy Image ◽  
Salt And Pepper Noise ◽  
Cortical Model ◽  
Gray Image ◽  
The Mean ◽  
Salt And Pepper

This paper introduces an efficient approach for feature extraction from noisy image using Intersecting Cortical Model(ICM), which is a simplified model of Pulse-Coupled Neural Network(PCNN). In our research, the entropy sequence of the output image, is obtained from the original gray image by ICM, as feature vector of the gray image, which can be used to represent the gray image, and this has been proved by our experiments. Consequently, it is used in the image classification, and the mean square error (MSE) between the feature vector of the input image and the standard feature vector is used to judge to which image groups the input image belongs. It has been proved that the method is not sensitivity with the Gaussian noise, salt and pepper noise or both of this and greatly robust for image recognition.

An Image Filtering Algorithm for Removing Mixed Noise

2013 ◽  
Vol 433-435 ◽  
pp. 383-388 ◽  
Author(s):  
Mao Xiang Chu ◽  
An Na Wang ◽  
Rong Fen Gong
Keyword(s):  
Gaussian Noise ◽  
Weighting Function ◽  
Median Filter ◽  
Bilateral Filter ◽  
Salt And Pepper Noise ◽  
Filtering Algorithm ◽  
Noise Simulation ◽  
Mixed Noise ◽  
Local Edge ◽  
Salt And Pepper

In order to remove salt-and-pepper noise and Gaussian noise in image, a novel filtering algorithm is proposed in this paper. The novel algorithm can preserve image edge details as much as possible. Firstly, five-median-binary code (FMBC) is proposed and used to describe local edge type of image. Secondly, median filter algorithm is improved to remove salt-and-pepper noise by using FMBC. Then, local enhanced bilateral filter with FMBC and a new type of exponential weighting function is used to remove Gaussian noise. Simulation results show that the algorithm proposed in this paper is very effective not only in filtering mixed noise but also in preserving edge details.

scholarly journals Nuclei Detection for Drug Discovery using Deep Learning

2019 ◽  
Vol 8 (2S8) ◽  
pp. 1289-1294
Keyword(s):  
Deep Learning ◽  
Drug Discovery ◽  
Discovery Process ◽  
Drug Discovery Process ◽  
Efficient System ◽  
Nuclei Detection ◽  
Starting Point ◽  
Net Framework ◽  
Deep Learning Model ◽  
And Training

Identifying a cell’s nucleus is the starting point for analysis of any kind of drug research. Presently this process is manually carried out by scientists. They take note of each nucleus from microscopic images to begin the drug discovery process. This takes hundreds of thousands of hours for scientific researchers to get their job done. In order to avoid such a bottleneck, this paper proposes an efficient solution using machine learning/ deep learning model. The proposed system can spot nuclei in cell images along with its run-length-encoded code without biologist intervention. A U-Net framework is used for the training the model to create efficient system. GPU based system is implemented to get accurate results for storage, retrieval and training of medical cell images. Thus, the system automates the spotting of nuclei thereby drastically reducing time in the drug discovery process.

Sign in / Sign up

Export Citation Format

Share Document