{"id":21297327,"url":"https://github.com/khrylx/dlow","last_synced_at":"2025-07-11T18:32:26.935Z","repository":{"id":124633412,"uuid":"280311095","full_name":"Khrylx/DLow","owner":"Khrylx","description":"[ECCV 2020] Official PyTorch Implementation of \"DLow: Diversifying Latent Flows for Diverse Human Motion Prediction\". ECCV 2020.","archived":false,"fork":false,"pushed_at":"2022-06-19T13:44:38.000Z","size":1666,"stargazers_count":119,"open_issues_count":1,"forks_count":18,"subscribers_count":4,"default_branch":"master","last_synced_at":"2025-04-06T07:09:14.601Z","etag":null,"topics":["deep-learning","diversifying-latent-flows","diversity","eccv2020","generative-models","motion-forecasting","motion-prediction","pose-forecasting","pose-prediction","pytorch","vae"],"latest_commit_sha":null,"homepage":"https://www.ye-yuan.com/dlow","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Khrylx.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null}},"created_at":"2020-07-17T02:52:03.000Z","updated_at":"2025-03-23T11:40:55.000Z","dependencies_parsed_at":null,"dependency_job_id":"0a8bd4bd-f6bd-41c6-a97c-813cf17dfe03","html_url":"https://github.com/Khrylx/DLow","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/Khrylx/DLow","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Khrylx%2FDLow","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Khrylx%2FDLow/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Khrylx%2FDLow/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Khrylx%2FDLow/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Khrylx","download_url":"https://codeload.github.com/Khrylx/DLow/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Khrylx%2FDLow/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":264870577,"owners_count":23676268,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","diversifying-latent-flows","diversity","eccv2020","generative-models","motion-forecasting","motion-prediction","pose-forecasting","pose-prediction","pytorch","vae"],"created_at":"2024-11-21T14:35:50.429Z","updated_at":"2025-07-11T18:32:26.003Z","avatar_url":"https://github.com/Khrylx.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# DLow: **D**iversifying **L**atent FL**ow**s\n![Loading DLow Overview](data/overview.png \"DLow Overview\")\n---\nThis repo contains the official implementation of our paper:\n  \nDLow: Diversifying Latent Flows for Diverse Human Motion Prediction  \nYe Yuan, Kris Kitani  \n**ECCV 2020**  \n[[website](https://www.ye-yuan.com/dlow)] [[paper](https://arxiv.org/pdf/2003.08386.pdf)] [[talk](https://youtu.be/c45ss6Tcb2A)] [[summary](https://youtu.be/nVYGHnRB1_M)] [[demo](https://youtu.be/64OEdSadb00)]\n\n# Installation \n### Datasets\n* Please follow the data preprocessing steps ([DATASETS.md](https://github.com/facebookresearch/VideoPose3D/blob/master/DATASETS.md)) inside the [VideoPose3D](https://github.com/facebookresearch/VideoPose3D) repo. Place the prepocessed data ``data_3d_h36m.npz`` (Human3.6M) and ``data_3d_humaneva15.npz`` (HumanEva-I) under the ``data`` folder.\n### Environment\n* **Tested OS:** MacOS, Linux\n* **Packages:**\n    * Python \u003e= 3.6\n    * [PyTorch](https://pytorch.org) \u003e= 0.4\n    * Tensorboard\n* **Note**: All scripts should be run from the root of this repo to avoid path issues.\n\n### Pretrained Models\n* Download our pretrained models from [Google Drive](https://drive.google.com/file/d/1k5uDeUXrvtwZPN-lJNPSO8tPvHH6Gj55/view?usp=sharing) (or [BaiduYun](https://pan.baidu.com/s/1Ye6bHXcX6lNVMLaXJyzyWg), password: y9ph) and place the unzipped ``results`` folder inside the root of this repo.\n\n# Train\n### Configs\nWe have provided 4 example YAML configs inside ``motion_pred/cfg``: \n* `h36m_nsamp10.yml` and `h36m_nsamp50.yml` for Human3.6M for number of samples 10 and 50 respectively.\n* `humaneva_nsamp10.yml` and `humaneva_nsamp50.yml` for HumanEva-I for number of samples 10 and 50 respectively.\n* These configs also have corresponding pretrained models inside ``results``.\n\n### Train VAE\n```\npython motion_pred/exp_vae.py --cfg h36m_nsamp10\n```\n\n### Train DLow (After VAE is trained)\n```\npython motion_pred/exp_dlow.py --cfg h36m_nsamp10\n```\n\n# Test \n### Visualize Motion Samples\n```\npython motion_pred/eval.py --cfg h36m_nsamp10 --mode vis\n```\nUseful keyboard shortcuts for the visualization GUI:  \n| Key           | Functionality |\n| ------------- | ------------- |\n| d             | test next motion data\n| c             | save current animation as `out/video.mp4` |\n| space         | stop/resume animation |\n| 1             | show DLow motion samples |\n| 2             | show VAE motion samples |\n\n\n### Compute Metrics\n```\npython motion_pred/eval.py --cfg h36m_nsamp50 --mode stats\n```  \n\n\n# Citation\nIf you find our work useful in your research, please cite our paper [DLow](https://www.ye-yuan.com/dlow):\n```bibtex\n@inproceedings{yuan2020dlow,\n    title={Dlow: Diversifying latent flows for diverse human motion prediction},\n    author={Yuan, Ye and Kitani, Kris},\n    booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},\n    year={2020}\n}\n```\n\n# Acknowledgement\nPart of the code is borrowed from the [VideoPose3D](https://github.com/facebookresearch/VideoPose3D) repo.\n\n# License\n\nThe software in this repo is freely available for free non-commercial use. Please see the [license](LICENSE) for further details.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkhrylx%2Fdlow","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fkhrylx%2Fdlow","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkhrylx%2Fdlow/lists"}