Description de l'offre
Intern,Deep Learning for Perceptual and Color-Aware Energy-Efficient Images à Rennes, France
InterDigital
- Temps pleinHoraires hebdomadaires standard
1 000 € a 1 500 €
Rennes, France
Summary The objective of this internship is to design a deep learning model for generating energy-aware images. Methods for producing energy-aware images — that is, images that consume less energy than their original versions when displayed on a screen — already exist in the literature. These approaches typically aim to achieve a trade-off between reducing energy consumption and preserving the visual quality of the resulting images. However, several research directions remain to be explored: Existing approaches mainly focus on reducing the luminance component of images, while we would also like to investigate the colour dimension. We would like to explore perceptual loss functions, i.e., losses designed to better reflect the characteristics of the human visual perception system. Current methods struggle to maintain a satisfactory quality-energy trade-off at high power reduction ratios. We would therefore like to investigate how the research directions above could improve performance in this more challenging range. The objective is to develop a model that combines these different aspects to improve both visual quality and energy efficiency compared with state-of-the-art approaches. This work represents a further step towards a more sustainable video streaming chain. In the context of climate change, reducing the environmental footprint of video streaming is an important part of the broader effort to reduce carbon emissions. The internship will first involve reviewing existing approaches for perceptual quality assessment and colour modelling of images. In a second stage, a deep learning-based model will be designed and implemented to generate energy-aware images using these new features. The performance of the proposed model will then be evaluated in terms of both image quality and energy consumption. The internship will take place within the Media Services research group at the InterDigital Video Lab. The intern will be mentored by research scientists and will contribute to a research project focused on developing solutions to reduce energy consumption across the video streaming ecosystem. Duration: 5-6 months, starting January-April 2026 Responsibilities State-of-the-art and analysis of existing solutions Implementation of a perceptually-based deep learning model for energy reduction Evaluation and reporting of results undefined Related work
- 1] O. Le Meur, C. -H. Demarty and L. Blonde, "Deep-Learning-Based Energy Aware Images," 2023 IEEE International Conference on Image Processing (ICIP), Kuala Lumpur, Malaysia, 2023, pp. 590-594, doi: 10.1109/ICIP49359.2.
- 2] Ameur, Z., Demarty, CH., Ménard, D., Meur, O.L. (2025). 3R-INN: How to Be Climate Friendly While Consuming/Delivering Videos?. In: Leonardis, A., Ricci, E., Roth, S., Russakovsky, O., Sattler, T., Varol, G. (eds) Computer Vision - ECCV 2024. ECCV 2024. Lecture Notes in Computer Science, vol 15133. Springer, Cham.
- 3] Emmanuel Sampaio, Claire-Hélène Demarty, and Olivier Le Meur. 2024. How to make images less power-hungry: An objective benchmark study. In Proceedings of the Second International ACM Green Multimedia Systems Workshop (GMSys '24). Association for Computing Machinery, New York, NY, USA, 1-7.
- 1] O. Le Meur, C. -H. Demarty and L. Blonde, "Deep-Learning-Based Energy Aware Images," 2023 IEEE International Conference on Image Processing (ICIP), Kuala Lumpur, Malaysia, 2023, pp. 590-594, doi: 10.1109/ICIP49359.2.
- 2] Ameur, Z., Demarty, CH., Ménard, D., Meur, O.L. (2025). 3R-INN: How to Be Climate Friendly While Consuming/Delivering Videos?. In: Leonardis, A., Ricci, E., Roth, S., Russakovsky, O., Sattler, T., Varol, G. (eds) Computer Vision - ECCV 2024. ECCV 2024. Lecture Notes in Computer Science, vol 15133. Springer, Cham.
- 3] Emmanuel Sampaio, Claire-Hélène Demarty, and Olivier Le Meur. 2024. How to make images less power-hungry: An objective benchmark study. In Proceedings of the Second International ACM Green Multimedia Systems Workshop (GMSys '24). Association for Computing Machinery, New York, NY, USA, 1-7.