{"id":1908,"date":"2025-06-22T00:31:52","date_gmt":"2025-06-22T00:31:52","guid":{"rendered":"https:\/\/catedramasmovil.uc3m.es\/2025\/06\/22\/upscaling-of-old-tv-cover-images\/"},"modified":"2025-06-22T00:52:31","modified_gmt":"2025-06-22T00:52:31","slug":"upscaling-of-old-tv-cover-images","status":"publish","type":"post","link":"https:\/\/catedramasmovil.uc3m.es\/en\/2025\/06\/22\/upscaling-of-old-tv-cover-images\/","title":{"rendered":"Upscaling of old TV cover images"},"content":{"rendered":"

[et_pb_section fb_built=”1″ _builder_version=”4.17.0″ custom_padding=”0px||||false|false” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_row _builder_version=”4.17.0″ _module_preset=”default” custom_padding=”0px||||false|false” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_column type=”4_4″ _builder_version=”4.17.0″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_gallery gallery_ids=”1898,1896,1894,1892,1890,1888,1886″ fullwidth=”on” _builder_version=”4.27.4″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”][\/et_pb_gallery][\/et_pb_column][\/et_pb_row][et_pb_row _builder_version=”4.16″ background_size=”initial” background_position=”top_left” background_repeat=”repeat” custom_margin=”|auto||103px||” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_column type=”4_4″ _builder_version=”4.16″ custom_padding=”|||” global_colors_info=”{}” custom_padding__hover=”|||” theme_builder_area=”post_content”][et_pb_text _builder_version=”4.25.1″ _module_preset=”default” custom_padding=”||0px|||” global_colors_info=”{}” theme_builder_area=”post_content”]<\/p>\n

Objective<\/h2>\n

This project focuses on the design and implementation of an advanced image super-resolution system to enhance the quality of vintage TV covers on the Orange TV platform. By leveraging state-of-the-art deep learning techniques (Real-ESRGAN, SwinFIR, HAT, etc.) alongside hyperparameter-optimization tools (Optuna), fine-tuning, and LoRA, it aims to automatically restore and upscale low-resolution images, balancing visual fidelity, perceived quality, and computational efficiency. <\/p>\n

This project focuses on the design and implementation of an advanced image super-resolution system to enhance the quality of vintage TV covers on the Orange TV platform. By leveraging state-of-the-art deep learning techniques (Real-ESRGAN, SwinFIR, HAT, etc.) alongside hyperparameter-optimization tools (Optuna), fine-tuning, and LoRA, it aims to automatically restore and upscale low-resolution images, balancing visual fidelity, perceived quality, and computational efficiency. <\/p>\n

The specific objectives are:<\/p>\n

\u25cf Compare super-resolution architectures (Real-ESRGAN, SwinFIR, HAT, etc.) by evaluating PSNR, SSIM, MS-SSIM, FSIM, and no-reference metrics (PIQE).<\/p>\n

\u25cf Optimize hyperparameters at inference with Optuna to maximize perceptual quality without sacrificing fidelity.<\/p>\n

\u25cf Fine-tune Real-ESRGAN on proprietary vintage cover data with typical distortions (resizing, noise, blur, compression).<\/p>\n

\u25cf Integrate LoRA to add low-rank adapter layers and speed up training iterations.<\/p>\n

\u25cf Define an evaluation protocol including inference-time analysis, standard deviation across multiple distortions, and scalability to large collections.<\/p>\n

\n

[\/et_pb_text][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=”1″ _builder_version=”4.18.0″ _module_preset=”default” custom_margin=”20px||||false|false” custom_padding=”0px||||false|false” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_row column_structure=”1_2,1_2″ _builder_version=”4.25.1″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_column type=”1_2″ _builder_version=”4.18.0″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_image src=”https:\/\/storage.googleapis.com\/wp-uploads.bucket.wp.uc3m.es\/wp-content\/uploads\/sites\/70\/2025\/06\/22002224\/imagen.jpg” title_text=”imagen” _builder_version=”4.27.4″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”][\/et_pb_image][\/et_pb_column][et_pb_column type=”1_2″ _builder_version=”4.18.0″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_text _builder_version=”4.27.4″ _module_preset=”default” custom_padding=”||0px|||” global_colors_info=”{}” theme_builder_area=”post_content”]<\/p>\n

BACHELOR’S THESIS BY:<\/strong><\/span><\/p>\n

NATALIA RODR\u00cdGUEZ NAVARRO<\/b><\/span><\/p>\n

[\/et_pb_text][et_pb_text _builder_version=”4.25.1″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”]<\/p>\n

<\/strong><\/p>\n

<\/strong><\/p>\n

<\/strong><\/p>\n

Academic Experience<\/strong><\/p>\n