{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/semantic-perceptual-image-compression-using","title":"Semantic Perceptual Image Compression using Deep Convolution Networks","arxiv_id":"1612.08712","date":"2016-12-27","proceeding":null,"authors":["Aaditya Prakash","Nick Moran","Solomon Garber","Antonella DiLillo","James Storer"],"abstract":"It has long been considered a significant problem to improve the visual\nquality of lossy image and video compression. Recent advances in computing\npower together with the availability of large training data sets has increased\ninterest in the application of deep learning cnns to address image recognition\nand image processing tasks. Here, we present a powerful cnn tailored to the\nspecific task of semantic image understanding to achieve higher visual quality\nin lossy compression. A modest increase in complexity is incorporated to the\nencoder which allows a standard, off-the-shelf jpeg decoder to be used. While\njpeg encoding may be optimized for generic images, the process is ultimately\nunaware of the specific content of the image to be compressed. Our technique\nmakes jpeg content-aware by designing and training a model to identify multiple\nsemantic regions in a given image. Unlike object detection techniques, our\nmodel does not require labeling of object positions and is able to identify\nobjects in a single pass. We present a new cnn architecture directed\nspecifically to image compression, which generates a map that highlights\nsemantically-salient regions so that they can be encoded at higher quality as\ncompared to background regions. By adding a complete set of features for every\nclass, and then taking a threshold over the sum of all feature activations, we\ngenerate a map that highlights semantically-salient regions so that they can be\nencoded at a better quality compared to background regions. Experiments are\npresented on the Kodak PhotoCD dataset and the MIT Saliency Benchmark dataset,\nin which our algorithm achieves higher visual quality for the same compressed\nsize.","url_abs":"http://arxiv.org/abs/1612.08712v2","url_pdf":"http://arxiv.org/pdf/1612.08712v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"semantic-perceptual-image-compression-using","repo_url":"https://github.com/iamaaditya/image-compression-cnn","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"semantic-perceptual-image-compression-using","repo_url":"https://github.com/anant95/K_means-image-compression","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"semantic-perceptual-image-compression-using","repo_url":"https://github.com/sakurusurya2000/CNN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"image-compression","task_name":"Image Compression"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"video-compression","task_name":"Video Compression"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}