IEEE Consumer Electronics Magazine - September/October 2020 - 100

Special Section on Machine Learning for End Consumers

preserving the quality of the visual content,
much compact images in both bitrate and resolution. The proposed approach outperforms the
BPG image codec.
In the current state, the proposed technique
focuses only on very low bitrate-compressed
images, but offers the potential to be extended
to a wider usage and for different applications.
This point is reinforced by the fact that perceptually driven models allow very good results for
automatic object recognition.10 In addition, the
BPG format is well-suited for critical CE fields
like IoT or healthcare.5
Perspectives of future work include the consideration of a more diversified training corpus
and the realization of a longer and better-adapted
ESRGAN training. In addition, we plan to extend
the model to different codecs (such as WebP),
under the perspective of a broader usage that
can overcome the BPG IP-related issues (patent
protection and royalties).

9. E. Agustsson, M. Tschannen, F. Mentzer, R. Timofte,
and L. V. Gool, "Generative adversarial networks for
extreme learned image compression," in Proc. IEEE
Int. Conf. Comput. Vision, 2019, pp. 221-231.
10. L. Galteri, L. Seidenari, M. Bertini, and A. Del Bimbo,
"Deep generative adversarial compression artifact
removal," in Proc. IEEE Int. Conf. Comput. Vision,
2017, pp. 4826-4835.
11. C. Dong, C. C. Loy, K. He, and X. Tang, "Image
super-resolution using deep convolutional networks,"
IEEE Trans. Pattern Anal. Mach. Intell., vol. 38, no. 2,
pp. 295-307, Feb. 2016.
12. Y. Romano, J. Isidoro, and P. Milanfar, "RAISR: Rapid
and accurate image super-resolution," IEEE Trans.
Comput. Imag., vol. 3, no. 1, pp. 110-125, Jan. 2017.
13. C. Ledig et al., "Photo-realistic single image
super-resolution using a generative adversarial
network," in Proc. IEEE Conf. Comput. Vis. Pattern
Recognit., 2017, pp. 4681-4690.
14. J. Nach, Saving you bandwidth through machine
learning, 2017. [Online]. Available: https://www.blog.
google/products/google-plus/saving-you-bandwidth-

& REFERENCES
1. T. Coughlin, "New media resolutions drive storage

through-machine-learning/
15. B. Lim, S. Son, H. Kim, S. Nah, and K. M. Lee, "Enhanced
deep residual networks for single image super-

requirements [The Art of Storage]," IEEE Consum.

resolution," in Proc. IEEE Conf. Comput. Vision Pattern

Electron. Mag., vol. 8, no. 1, pp. 90-91, Jan. 2019.

Recognit. Workshops, vol. 1, no. 2, pp. 1132-1140,

2. T. Coughlin, "When memory starts to think [The Art of
Storage]," IEEE Consum. Electron. Mag., vol. 8, no. 3,
pp. 90-91, May 2019.
3. V. Bucur, C. Dehelean, and L. Miclea, "Object storage
in the cloud and multi-cloud: State of the art and the

Jul. 2017.
16. X. Wang et al., "ESRGAN: Enhanced super-resolution
generative adversarial networks," in Proc. IEEE Int.
Conf. Comput. Vision, Sep. 2018, pp. 63-79.
17. S. Chikkerur, V. Sundaram, M. Reisslein, and

research challenges," in Proc. IEEE Int. Conf.

L. J. Karam, "Objective video quality assessment

Automat., Qual. Test., Robot., May 2018, pp. 1-6.

methods: A classification, review, and performance

4. F. Bellard, BPG Image format, 2015. [Online].
Available: https://bellard.org/bpg/
5. S. P. Mohanty, E. Kougianos, and P. Guturu, "SBPG:

comparison," IEEE Trans. Broadcast., vol. 57, no. 2,
pp. 165-182, Jun. 2011.
18. Recommendation ITU-R BT.500-13, Methodology for

Secure better portable graphics for trustworthy media

the subjective assessment of the quality of television

communications in the IoT," IEEE Access, vol. 6,

pictures, Jan. 2012. [Online]. Available: https://www.

pp. 5939-5953, 2018.

itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.500-13-

6. I. Goodfellow et al., "Generative adversarial nets," in Proc.

201201-I!!PDF-E.pdf

Adv. Neural Inf. Process. Syst., 2014, pp. 2672-2680.
7. J. Johnson, A. Alahi, and L. Fei-Fei, "Perceptual losses
for real-time style transfer and super-resolution," in Proc.
Eur. Conf. Comput. Vision, Oct. 2016, pp. 694-711.
8. K. Simonyan and A. Zisserman, "Very deep
convolutional networks for large-scale image
recognition," in Proc. 3rd Int. Conf. Learn.
Representations, San Diego, CA, USA, 2015.

100

 bastien Hamis is currently working toward
Se
the Ph.D. degree with the ARTEMIS Department of
 le
com SudParis, Paris, France. His research interTe
ests are focus on image compression, superresolution, and more generally on deep learning applied
to image enhancement. Contact him at sebastien.
hamis@telecom-sudparis.eu.

IEEE Consumer Electronics Magazine

https://www.blog.google/products/google-plus/saving-you-bandwidth-through-machine-learning/ https://www.blog.google/products/google-plus/saving-you-bandwidth-through-machine-learning/ https://www.blog.google/products/google-plus/saving-you-bandwidth-through-machine-learning/ https://www.bellard.org/bpg/ https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.500-13-201201-I!!PDF-E.pdf https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.500-13-201201-I!!PDF-E.pdf https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.500-13-201201-I!!PDF-E.pdf

IEEE Consumer Electronics Magazine - September/October 2020

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - September/October 2020