{"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/facial-aging-and-rejuvenation-by-conditional","title":"Facial Aging and Rejuvenation by Conditional Multi-Adversarial Autoencoder with Ordinal Regression","arxiv_id":"1804.02740","date":"2018-04-08","proceeding":null,"authors":["Haiping Zhu","Qi Zhou","Junping Zhang","James Z. Wang"],"abstract":"Facial aging and facial rejuvenation analyze a given face photograph to\npredict a future look or estimate a past look of the person. To achieve this,\nit is critical to preserve human identity and the corresponding aging\nprogression and regression with high accuracy. However, existing methods cannot\nsimultaneously handle these two objectives well. We propose a novel generative\nadversarial network based approach, named the Conditional Multi-Adversarial\nAutoEncoder with Ordinal Regression (CMAAE-OR). It utilizes an age estimation\ntechnique to control the aging accuracy and takes a high-level feature\nrepresentation to preserve personalized identity. Specifically, the face is\nfirst mapped to a latent vector through a convolutional encoder. The latent\nvector is then projected onto the face manifold conditional on the age through\na deconvolutional generator. The latent vector preserves personalized face\nfeatures and the age controls facial aging and rejuvenation. A discriminator\nand an ordinal regression are imposed on the encoder and the generator in\ntandem, making the generated face images to be more photorealistic while\nsimultaneously exhibiting desirable aging effects. Besides, a high-level\nfeature representation is utilized to preserve personalized identity of the\ngenerated face. Experiments on two benchmark datasets demonstrate appealing\nperformance of the proposed method over the state-of-the-art.","url_abs":"http://arxiv.org/abs/1804.02740v1","url_pdf":"http://arxiv.org/pdf/1804.02740v1.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":[],"tasks":[{"task_slug":"age-estimation","task_name":"Age Estimation"},{"task_slug":null,"task_name":"Generative Adversarial Network"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/age-estimation-on-fgnet","task":"Age Estimation","dataset":"FGNET","model":"CMAAE-OR","rank_in_archive_order":5,"of":8,"metrics":{"MAE":"3.62"},"uses_additional_data":false},{"leaderboard":"/sota/age-estimation-on-fgnet","task":"Age Estimation","dataset":"FGNET","model":"Zhu et al. (Actual)","rank_in_archive_order":7,"of":8,"metrics":{"MAE":"4.58"},"uses_additional_data":false},{"leaderboard":"/sota/age-estimation-on-morph","task":"Age Estimation","dataset":"MORPH","model":"CMAAE-OR","rank_in_archive_order":1,"of":1,"metrics":{"MAE":"1.48"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}