{"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/futuredepth-learning-to-predict-the-future","title":"FutureDepth: Learning to Predict the Future Improves Video Depth Estimation","arxiv_id":"2403.12953","date":"2024-03-19","proceeding":null,"authors":["Rajeev Yasarla","Manish Kumar Singh","Hong Cai","Yunxiao Shi","Jisoo Jeong","Yinhao Zhu","Shizhong Han","Risheek Garrepalli","Fatih Porikli"],"abstract":"In this paper, we propose a novel video depth estimation approach, FutureDepth, which enables the model to implicitly leverage multi-frame and motion cues to improve depth estimation by making it learn to predict the future at training. More specifically, we propose a future prediction network, F-Net, which takes the features of multiple consecutive frames and is trained to predict multi-frame features one time step ahead iteratively. In this way, F-Net learns the underlying motion and correspondence information, and we incorporate its features into the depth decoding process. Additionally, to enrich the learning of multiframe correspondence cues, we further leverage a reconstruction network, R-Net, which is trained via adaptively masked auto-encoding of multiframe feature volumes. At inference time, both F-Net and R-Net are used to produce queries to work with the depth decoder, as well as a final refinement network. Through extensive experiments on several benchmarks, i.e., NYUDv2, KITTI, DDAD, and Sintel, which cover indoor, driving, and open-domain scenarios, we show that FutureDepth significantly improves upon baseline models, outperforms existing video depth estimation methods, and sets new state-of-the-art (SOTA) accuracy. Furthermore, FutureDepth is more efficient than existing SOTA video depth estimation models and has similar latencies when comparing to monocular models","url_abs":"https://arxiv.org/abs/2403.12953v2","url_pdf":"https://arxiv.org/pdf/2403.12953v2.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":"decoder","task_name":"Decoder"},{"task_slug":"depth-estimation","task_name":"Depth Estimation"},{"task_slug":"future-prediction","task_name":"Future prediction"},{"task_slug":"monocular-depth-estimation","task_name":"Monocular Depth Estimation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/monocular-depth-estimation-on-kitti-eigen","task":"Monocular Depth Estimation","dataset":"KITTI Eigen split","model":"FutureDepth","rank_in_archive_order":6,"of":79,"metrics":{"Delta < 1.25":"0.984","Delta < 1.25^2":"0.998","Delta < 1.25^3":"1.000","RMSE":"1.856","RMSE log":"0.066","Sq Rel":"0.117","Square relative error (SqRel)":"0.117","absolute relative error":"0.041"},"uses_additional_data":false},{"leaderboard":"/sota/monocular-depth-estimation-on-nyu-depth-v2","task":"Monocular Depth Estimation","dataset":"NYU-Depth V2","model":"FutureDepth","rank_in_archive_order":18,"of":85,"metrics":{"Delta < 1.25":"0.981","Delta < 1.25^2":"0.996","Delta < 1.25^3":"0.999","RMSE":"0.233","absolute relative error":"0.063","log 10":"0.027"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2403.12953","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}