{"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/densely-connected-pyramid-dehazing-network","title":"Densely Connected Pyramid Dehazing Network","arxiv_id":"1803.08396","date":"2018-03-22","proceeding":"CVPR 2018 6","authors":["He Zhang","Vishal M. Patel"],"abstract":"We propose a new end-to-end single image dehazing method, called Densely\nConnected Pyramid Dehazing Network (DCPDN), which can jointly learn the\ntransmission map, atmospheric light and dehazing all together. The end-to-end\nlearning is achieved by directly embedding the atmospheric scattering model\ninto the network, thereby ensuring that the proposed method strictly follows\nthe physics-driven scattering model for dehazing. Inspired by the dense network\nthat can maximize the information flow along features from different levels, we\npropose a new edge-preserving densely connected encoder-decoder structure with\nmulti-level pyramid pooling module for estimating the transmission map. This\nnetwork is optimized using a newly introduced edge-preserving loss function. To\nfurther incorporate the mutual structural information between the estimated\ntransmission map and the dehazed result, we propose a joint-discriminator based\non generative adversarial network framework to decide whether the corresponding\ndehazed image and the estimated transmission map are real or fake. An ablation\nstudy is conducted to demonstrate the effectiveness of each module evaluated at\nboth estimated transmission map and dehazed result. Extensive experiments\ndemonstrate that the proposed method achieves significant improvements over the\nstate-of-the-art methods. Code will be made available at:\nhttps://github.com/hezhangsprinter","url_abs":"http://arxiv.org/abs/1803.08396v1","url_pdf":"http://arxiv.org/pdf/1803.08396v1.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":"densely-connected-pyramid-dehazing-network","repo_url":"https://github.com/hezhangsprinter/DCPDN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":null,"task_name":"Generative Adversarial Network"},{"task_slug":"image-dehazing","task_name":"Image Dehazing"},{"task_slug":"single-image-dehazing","task_name":"Single Image Dehazing"}],"methods":[{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"pyramid-pooling-module","method_name":"Pyramid Pooling Module"},{"method_slug":"relu","method_name":"ReLU"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-dehazing-on-reside-6k","task":"Image Dehazing","dataset":"RESIDE-6K","model":"DCPDN","rank_in_archive_order":6,"of":6,"metrics":{"PSNR":"19.39","SSIM":"0.65"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1803.08396","atlas_url":"https://app.syntology.ai/?focus=1803.08396","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}