{"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/towards-real-time-unsupervised-monocular","title":"Towards real-time unsupervised monocular depth estimation on CPU","arxiv_id":"1806.11430","date":"2018-06-29","proceeding":null,"authors":["Matteo Poggi","Filippo Aleotti","Fabio Tosi","Stefano Mattoccia"],"abstract":"Unsupervised depth estimation from a single image is a very attractive\ntechnique with several implications in robotic, autonomous navigation,\naugmented reality and so on. This topic represents a very challenging task and\nthe advent of deep learning enabled to tackle this problem with excellent\nresults. However, these architectures are extremely deep and complex. Thus,\nreal-time performance can be achieved only by leveraging power-hungry GPUs that\ndo not allow to infer depth maps in application fields characterized by\nlow-power constraints. To tackle this issue, in this paper we propose a novel\narchitecture capable to quickly infer an accurate depth map on a CPU, even of\nan embedded system, using a pyramid of features extracted from a single input\nimage. Similarly to state-of-the-art, we train our network in an unsupervised\nmanner casting depth estimation as an image reconstruction problem. Extensive\nexperimental results on the KITTI dataset show that compared to the top\nperforming approach our network has similar accuracy but a much lower\ncomplexity (about 6% of parameters) enabling to infer a depth map for a KITTI\nimage in about 1.7 s on the Raspberry Pi 3 and at more than 8 Hz on a standard\nCPU. Moreover, by trading accuracy for efficiency, our network allows to infer\nmaps at about 2 Hz and 40 Hz respectively, still being more accurate than most\nstate-of-the-art slower methods. To the best of our knowledge, it is the first\nmethod enabling such performance on CPUs paving the way for effective\ndeployment of unsupervised monocular depth estimation even on embedded systems.","url_abs":"http://arxiv.org/abs/1806.11430v3","url_pdf":"http://arxiv.org/pdf/1806.11430v3.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":"towards-real-time-unsupervised-monocular","repo_url":"https://github.com/mattpoggi/pydnet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"towards-real-time-unsupervised-monocular","repo_url":"https://github.com/FilippoAleotti/mobilePydnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"towards-real-time-unsupervised-monocular","repo_url":"https://github.com/Pomu0708/PyDnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"towards-real-time-unsupervised-monocular","repo_url":"https://github.com/dronefreak/dji-tello-collision-avoidance-pydnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"autonomous-navigation","task_name":"Autonomous Navigation"},{"task_slug":null,"task_name":"CPU"},{"task_slug":"depth-estimation","task_name":"Depth Estimation"},{"task_slug":"image-reconstruction","task_name":"Image Reconstruction"},{"task_slug":"monocular-depth-estimation","task_name":"Monocular Depth Estimation"},{"task_slug":"raspberry-pi-3","task_name":"Raspberry Pi 3"},{"task_slug":"unsupervised-monocular-depth-estimation","task_name":"Unsupervised Monocular Depth Estimation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.11430","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}