Deep Colorization

Cheng, Zezhou; Yang, Qingxiong; Sheng, Bin

doi:10.1109/iccv.2015.55

Cited by 521 publications

(319 citation statements)

References 39 publications

Supporting

Mentioning

316

Contrasting

Unclassified

Order By: Relevance

“…It can be seen that we could produce more realistic and plausible colors than most state-of-the-art example-based colorization algorithms. Figure 9 presents the Peak Signal-to-Noise Ratio (PSNR) distribution of our method, Cheng et al [17], and Deshpande et al [16]. We have measured the PSNR distribution on 1500 test images from the SUN database [32].…”

Section: Resultsmentioning

confidence: 99%

“…Deshpande et al [16] colorize an image by optimizing a linear system that considers local predictions of color, spatial consistency, and consistency with an overall histogram. Cheng et al [17] introduced a fully-automatic method based on a deep neural network which was trained by hand-crafted features. Three levels of features were extracted from each pixel of the training images: raw grayscale values, DAISY features [18], and high-level semantic features.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Twin Deep Convolutional Neural Network for Example-Based Image Colorization

Varga

Szirányi

2017

Computer Analysis of Images and Patterns

View full text Add to dashboard Cite

Abstract. This paper deals with the colorization of grayscale images. Recent papers have shown remarkable results on image colorization utilizing various deep architectures. Unlike previous methods, we perform colorization using a deep architecture and a reference image. Our architecture utilizes two parallel Convolutional Neural Networks which have the same structure. One CNN, which uses the reference image, helps the other CNN in color prediction for the input image. On the other hand, the second CNN, which uses the input image, helps to identify the areas which holds essential information about the color scheme of the scene. Comprehensive experiments and qualitative and quantitative evaluations were conducted on the images of SUN database and on other images. Quantitative evaluations are based on Peak Signal-to-Noise Ratio (PSNR) and on Quaternion Structural Similarity (QSSIM).

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Twin Deep Convolutional Neural Network for Example-Based Image Colorization

Varga

Szirányi

2017

Computer Analysis of Images and Patterns

View full text Add to dashboard Cite

show abstract

“…This approach, minimization of the least squares, was also employed in deep learning application for various types of image processing (Cheng et al, 2015;Dong et al, 2014). Furthermore, 3-dimensional MR images need different methods from recent segmentation methods of 2-dimensional images (e.g.…”

Section: Discussionmentioning

confidence: 99%

Fast and robust segmentation of the striatum using deep convolutional neural networks

Choi¹,

Jin

2016

Journal of Neuroscience Methods

View full text Add to dashboard Cite

h i g h l i g h t s• We describe a new method for stratium segmentation.• We employ two serial deep convolutional neural networks (CNN).• Segmentation accuracy of deep CNN is comparable with that of previous methods.• The processing time of the new method is much faster than previous methods. a r t i c l e i n f o a b s t r a c tBackground: Automated segmentation of brain structures is an important task in structural and functional image analysis. We developed a fast and accurate method for the striatum segmentation using deep convolutional neural networks (CNN). New method: T1 magnetic resonance (MR) images were used for our CNN-based segmentation, which require neither image feature extraction nor nonlinear transformation. We employed two serial CNN, Global and Local CNN: The Global CNN determined approximate locations of the striatum. It performed a regression of input MR images fitted to smoothed segmentation maps of the striatum. From the output volume of Global CNN, cropped MR volumes which included the striatum were extracted. The cropped MR volumes and the output volumes of Global CNN were used for inputs of Local CNN. Local CNN predicted the accurate label of all voxels. Segmentation results were compared with a widely used segmentation method, FreeSurfer. Results: Our method showed higher Dice Similarity Coefficient (DSC) (0.893 ± 0.017 vs. 0.786 ± 0.015) and precision score (0.905 ± 0.018 vs. 0.690 ± 0.022) than FreeSurfer-based striatum segmentation (p = 0.06). Our approach was also tested using another independent dataset, which showed high DSC (0.826 ± 0.038) comparable with that of FreeSurfer.Comparison with existing method Segmentation performance of our proposed method was comparable with that of FreeSurfer. The running time of our approach was approximately three seconds. Conclusion: We suggested a fast and accurate deep CNN-based segmentation for small brain structures which can be widely applied to brain image analysis.

show abstract

“…It have been demonstrated to be very effective in various computer vision and image processing tasks including pedestrian detection [4], face detection [5], image classification [6], image super-resolution [7], automatic image colorization [8] etc. CNN-based methods have been applied on the task of image retrieval recently.…”

Section: Related Workmentioning

confidence: 99%

Fast content-based image retrieval using convolutional neural network and hash function

Varga

Szirányi

2016

2016 IEEE International Conference on Systems, Man, and Cybernetics (SMC)

View full text Add to dashboard Cite

Abstract-Due to the explosive increase of online images, content-based image retrieval has gained a lot of attention. The success of deep learning techniques such as convolutional neural networks have motivated us to explore its applications in our context. The main contribution of our work is a novel endto-end supervised learning framework that learns probabilitybased semantic-level similarity and feature-level similarity simultaneously. The main advantage of our novel hashing scheme that it is able to reduce the computational cost of retrieval significantly at the state-of-the-art efficiency level. We report on comprehensive experiments using public available datasets such as Oxford, Holidays and ImageNet 2012 retrieval datasets.

show abstract

Deep Colorization

Cited by 521 publications

References 39 publications

Twin Deep Convolutional Neural Network for Example-Based Image Colorization

Twin Deep Convolutional Neural Network for Example-Based Image Colorization

Fast and robust segmentation of the striatum using deep convolutional neural networks

Fast content-based image retrieval using convolutional neural network and hash function

Contact Info

Product

Resources

About