Karlo Koščević
ABSTRACT
Deep learning methods have achieved a large step forward in many computer vision applications. With mechanisms such as attention, deep models can now guide themselves to focus on parts of an image that are more significant for a given task. In computational color constancy, the most important step is to estimate the illumination vector as accurately as possible. Since illumination estimation algorithms can be sensitive to noise, such as ambiguous regions in the image, the ability to have a mechanism to look for specific regions in an image could be helpful. In this paper, a convolutional neural network with an attention mechanism is proposed. The attention mechanism helps the network to focus on regions that contain more content and to avoid regions where ambiguous estimations may occur. In the experimental results, it is shown that the attention mechanism does help the network to obtain more accurate estimations and puts the focus of the network on the regions in an image where gradients are high. The network with the attention mechanism achieves up to 10% increase in accuracy compared to the same network architecture without the attention mechanism.
- A. Akbarinia and C. A. Parraga. 2018. Colour Constancy Beyond the Classical Receptive Field. IEEE Transactions on Pattern Analysis and Machine Intelligence 40, 9 (Sep. 2018), 2081--2094. https://doi.org/10.1109/TPAMI.2017.2753239Google Scholar
Digital Library
- Nikola Banić and Sven Lončarić. 2013. Using the Random Sprays Retinex Algorithm for Global Illumination Estimation. In Proceedings of The Second Croatian Computer Vision Workshopn (CCVW 2013). University of Zagreb Faculty of Electrical Engineering and Computing, 3--7.Google Scholar
- Nikola Banić and Sven Lončarić. 2014. Color Rabbit: Guiding the Distance of Local Maximums in Illumination Estimation. In Digital Signal Processing (DSP), 2014 19th International Conference on. IEEE, 345--350.Google ScholarCross Ref
- Nikola Banić and Sven Lončarić. 2014. Improving the White patch method by subsampling. In Image Processing (ICIP), 2014 21st IEEE International Conference on. IEEE, 605--609.Google ScholarCross Ref
- Nikola Banić and Sven Lončarić. 2015. A Perceptual Measure of Illumination Estimation Error. In VISAPP. 136--143.Google Scholar
- Nikola Banić and Sven Lončarić. 2015. Color Cat: Remembering Colors for Illumination Estimation. Signal Processing Letters, IEEE 22, 6 (2015), 651--655.Google Scholar
- Nikola Banić and Sven Lončarić. 2015. Color Dog: Guiding the Global Illumination Estimation to Better Accuracy. In VISAPP. 129--135.Google Scholar
- Nikola Banić and Sven Lončarić. 2015. Using the red chromaticity for illumination estimation. In Image and Signal Processing and Analysis (ISPA), 2015 9th International Symposium on. IEEE, 131--136.Google ScholarCross Ref
- Nikola Banić and Sven Lončarić. 2018. Green Stability Assumption: Unsupervised Learning for Statistics-Based Illumination Estimation. Journal of Imaging 4, 11 (2018), 127.Google ScholarCross Ref
- Nikola Banić and Sven Lončarić. 2018. Unsupervised Learning for Color Constancy. In VISAPP. 181--188.Google Scholar
- Nikola Banić and Sven Lončarić. 2019. Blue Shift Assumption: Improving Illumination Estimation Accuracy for Single Image from Unknown Source. In VISAPP. 191--197.Google Scholar
- Kobus Barnard, Vlad Cardei, and Brian Funt. 2002. A comparison of computational color constancy algorithms. I: Methodology and experiments with synthesized data. Image Processing, IEEE Transactions on 11, 9 (2002), 972--984.Google ScholarDigital Library
- Jonathan T Barron. 2015. Convolutional Color Constancy. In Proceedings of the IEEE International Conference on Computer Vision. 379--387.Google ScholarDigital Library
- Jonathan T Barron and Yun-Ta Tsai. 2017. Fast Fourier Color Constancy. In Computer Vision and Pattern Recognition, 2017. CVPR 2017. IEEE Computer Society Conference on, Vol. 1. IEEE.Google ScholarCross Ref
- Simone Bianco, Claudio Cusano, and Raimondo Schettini. 2015. Color Constancy Using CNNs. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops. 81--89.Google ScholarCross Ref
- Gershon Buchsbaum. 1980. A spatial processor model for object colour perception. Journal of The Franklin Institute 310, 1 (1980), 1--26.Google ScholarCross Ref
- Vlad C Cardei, Brian Funt, and Kobus Barnard. 2002. Estimating the scene illumination chromaticity by using a neural network. JOSA A 19, 12 (2002), 2374--2386.Google ScholarCross Ref
- Ayan Chakrabarti, Keigo Hirakawa, and Todd Zickler. 2012. Color constancy with spatio-spectral statistics. Pattern Analysis and Machine Intelligence, IEEE Transactions on 34, 8 (2012), 1509--1519.Google ScholarDigital Library
- Dongliang Cheng, Dilip K Prasad, and Michael S Brown. 2014. Illuminant estimation for color constancy: why spatial-domain methods work and the role of the color distribution. JOSA A 31, 5 (2014), 1049--1058.Google ScholarCross Ref
- Dongliang Cheng, Brian Price, Scott Cohen, and Michael S Brown. 2015. Effective learning-based illuminant estimation using simple features. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 1000--1008.Google ScholarCross Ref
- Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. 2009. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition. Ieee, 248--255.Google ScholarCross Ref
- Marc Ebner. 2007. Color Constancy. Wiley.Google Scholar
- Graham D Finlayson. 2013. Corrected-moment illuminant estimation. In Proceedings of the IEEE International Conference on Computer Vision. 1904--1911.Google ScholarDigital Library
- Graham D Finlayson, Mark S Drew, and Brian V Funt. 1994. Color constancy: generalized diagonal transforms suffice. JOSA A 11, 11 (1994), 3011--3019.Google ScholarCross Ref
- Graham D Finlayson and Elisabetta Trezzi. 2004. Shades of gray and colour constancy. In Color and Imaging Conference, Vol. 2004. Society for Imaging Science and Technology, 37--41.Google ScholarCross Ref
- Brian Funt and Lilong Shi. 2010. The rehabilitation of MaxRGB. In Color and Imaging Conference, Vol. 2010. Society for Imaging Science and Technology, 256--259.Google Scholar
- Shao-Bing Gao, Kai-Fu Yang, Chao-Yi Li, and Yong-Jie Li. 2015. Color constancy using double-opponency. IEEE Transactions on Pattern Analysis and Machine Intelligence 37, 10 (2015), 1973--1985.Google ScholarDigital Library
- Peter V Gehler, Carsten Rother, Andrew Blake, Tom Minka, and Toby Sharp. 2008. Bayesian color constancy revisited. In Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference on. IEEE, 1--8.Google ScholarCross Ref
- Arjan Gijsenij and Theo Gevers. 2007. Color Constancy using Natural Image Statistics.. In CVPR. 1--8.Google Scholar
- Arjan Gijsenij, Theo Gevers, and Joost Van DeWeijer. 2011. Computational color constancy: Survey and experiments. Image Processing, IEEE Transactions on 20, 9 (2011), 2475--2489.Google ScholarDigital Library
- Arjan Gijsenij, Theo Gevers, and Joost Van DeWeijer. 2012. Improving color constancy by photometric edge weighting. Pattern Analysis and Machine Intelligence, IEEE Transactions on 34, 5 (2012), 918--929.Google ScholarDigital Library
- Steven D Hordley. 2006. Scene illuminant estimation: past, present, and future. Color Research & Application: Endorsed by Inter-Society Color Council, The Colour Group (Great Britain), Canadian Society for Color, Color Science Association of Japan, Dutch Society for the Study of Color, The Swedish Colour Centre Foundation, Colour Society of Australia, Centre Français de la Couleur 31, 4 (2006), 303--314.Google Scholar
- Yuanming Hu, BaoyuanWang, and Stephen Lin. 2017. Fully Convolutional Color Constancy with Confidence-weighted Pooling. In Computer Vision and Pattern Recognition, 2017. CVPR 2017. IEEE Conference on. IEEE, 4085--4094.Google ScholarCross Ref
- Hamid Reza Vaezi Joze and Mark S Drew. 2013. Exemplar-based color constancy and multiple illumination. IEEE transactions on pattern analysis and machine intelligence 36, 5 (2013), 860--873.Google Scholar
- Hamid Reza Vaezi Joze, Mark S Drew, GrahamDFinlayson, and PerlaAurora Troncoso Rey. 2012. The Role of Bright Pixels in Illumination Estimation. In Color and Imaging Conference, Vol. 2012. Society for Imaging Science and Technology, 41--46.Google Scholar
- Karlo Koščević, Nikola Banić, and Sven Lončarić. 2019. Color Beaver: Bounding Illumination Estimations for Higher Accuracy. In VISAPP. 183--190.Google Scholar
- Karlo Koščević, Marko Subašić, and Sven Lončarić. 2019. Attention-based Convolutional Neural Network for Computer Vision Color Constancy. In 2019 11th International Symposium on Image and Signal Processing and Analysis (ISPA). IEEE, 372--377.Google Scholar
- Edwin H Land. 1977. The retinex theory of color vision. Scientific America.Google Scholar
- Kunpeng Li, Ziyan Wu, Kuan-Chuan Peng, Jan Ernst, and Yun Fu. 2018. Tell me where to look: Guided attention inference network. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 9215--9223.Google ScholarCross Ref
- Seoung Wug Oh and Seon Joo Kim. 2017. Approaching the computational color constancy as a classification problem through deep learning. Pattern Recognition 61 (2017), 405--416.Google ScholarCross Ref
- Yanlin Qian, Said Pertuz, Jarno Nikkanen, Joni-Kristian K am ar ainen, and Jiri Matas. 2019. Revisiting Gray Pixel for Statistical Illumination Estimation. In VISAPP. 36--46.Google Scholar
- Wu Shi, Chen Change Loy, and Xiaoou Tang. 2016. Deep Specialized Network for Illuminant Estimation. In European Conference on Computer Vision. Springer, 371--387.Google Scholar
- Oleksii Sidorov. 2018. Artificial Color Constancy via GoogLeNet with Angular Loss Function. arXiv preprint arXiv:1811.08456 (2018).Google Scholar
- Karen Simonyan and AndrewZisserman. 2014. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014).Google Scholar
- Joost Van De Weijer, Theo Gevers, and Arjan Gijsenij. 2007. Edge-based color constancy. Image Processing, IEEE Transactions on 16, 9 (2007), 2207--2214.Google ScholarDigital Library
- Joost Van De Weijer, Cordelia Schmid, and Jakob Verbeek. 2007. Using high-level visual information for color constancy. In Computer Vision, 2007. ICCV 2007. IEEE 11th International Conference on. IEEE, 1--8.Google ScholarCross Ref
- Javier Vazquez-Corral, Maria Vanrell, Ramon Baldrich, and Francesc Tous. 2011. Color constancy by category correlation. IEEE Transactions on image processing 21, 4 (2011), 1997-2007.Google ScholarDigital Library
- S.Woo, S. Lee, J. Yoo, and J. Kim. 2018. Improving Color Constancy in an Ambient Light Environment Using the Phong Reflection Model. IEEE Transactions on Image Processing 27, 4 (April 2018), 1862-1877. https://doi.org/10.1109/TIP.2017.2785290Google ScholarDigital Library
- Kai-Fu Yang, Shao-Bing Gao, and Yong-Jie Li. 2015. Efficient illuminant estimation for color constancy using grey pixels. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2254--2263.Google ScholarCross Ref
Index Terms
- Guiding the Illumination Estimation Using the Attention Mechanism
Recommendations
Retinex low-light image enhancement network based on attention mechanism
AbstractImages taken at night, under cloudy conditions, etc., are degraded and of poor quality, which can cause problems such as low brightness, excessive noise, and loss of information. In recent years, deep learning-based methods have led to a ...
A grayscale image enhancement algorithm based on dense residual and attention mechanism
AbstractDeep learning shows great potential in low-light image enhancement, which can improve image brightness and contrast while keeping image natural. However, due to the lack of prior knowledge extracted manually or excessive amplification of noise, ...
Low-light image enhancement based on GAN with attention mechanism and color Constancy
AbstractImages captured in low-light often suffer from severe quality degraded problems, such as low contrast and color distortion, which make it intractable for further computer vision tasks. To solve the problems above, we proposed a trainable parallel ...
Comments