{"id":22230358,"url":"https://github.com/line/ChronAccRet","last_synced_at":"2025-07-27T19:32:50.898Z","repository":{"id":251860584,"uuid":"838116025","full_name":"line/ChronAccRet","owner":"line","description":null,"archived":false,"fork":false,"pushed_at":"2024-08-05T14:29:22.000Z","size":157,"stargazers_count":0,"open_issues_count":0,"forks_count":1,"subscribers_count":5,"default_branch":"master","last_synced_at":"2024-08-06T07:32:12.867Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/line.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":"CODE_OF_CONDUCT.md","threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2024-08-05T01:18:03.000Z","updated_at":"2024-08-06T07:32:22.902Z","dependencies_parsed_at":"2024-08-06T07:42:21.515Z","dependency_job_id":null,"html_url":"https://github.com/line/ChronAccRet","commit_stats":null,"previous_names":["line/chronaccret"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/line%2FChronAccRet","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/line%2FChronAccRet/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/line%2FChronAccRet/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/line%2FChronAccRet/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/line","download_url":"https://codeload.github.com/line/ChronAccRet/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":227831080,"owners_count":17826155,"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":[],"created_at":"2024-12-03T01:14:54.948Z","updated_at":"2024-12-03T01:15:00.817Z","avatar_url":"https://github.com/line.png","language":"Python","readme":"#  Official Implementation of Chronologically Accurate Retrieval for Temporal Grounding of Motion-Language Models (ECCV 2024)\n\nOfficial implementation of the paper: Chronologically Accurate Retrieval for Temporal Grounding of Motion-Language Models.\n\n## Framework\n\n![Framework](./assets/figure.jpg)\n\n## Requirements\n\nPython 3.10\n\nPytorch 2.0.0\n\n### Recommended: Using Poetry\n\n```bash\npoetry install\n```\n\n### Using Conda\n```bash\nconda create -n car python=3.11\nconda activate car\nconda install pytorch==2.0.0 torchvision==0.15.1 torchaudio==2.0.1 pytorch-cuda=11.8 -c pytorch -c nvidia\npip install -r requirements.txt\n```\n\n## Data Preparation\nPlease follow the descriptions in the [TMR](https://github.com/Mathux/TMR) repository to prepare the datasets.\nThen, place them as follows.\nNote that events have to be generated as described in the paper. Please prepare a json file for each motion, with \"events\" indicating the decomposed events. You can download the event descriptions from [here](https://drive.google.com/file/d/14FUVBotloBLDbtpvlShQTyX5A4TogLL4/view?usp=drive_link).\nPlease view \"datasets/datasets.py\" for how the files are handled.\n\n\n```\nChronAccRet\n│   README.md\n│   requirements.txt\n|   ...\n|\n└───config\n└───datasets\n└───...\n└───stats\n    └───humanml3d\n        └───guoh3dfeats #place the statistics here\n            └───mean.pt\n                std.pt\n└───data\n    └───humanml3d\n        │   annotations.json\n        └───event_texts  #texts for events (the json files described above)\n        └───motions\n        └───sent_embeddings #embeddings of sentences\n        └───splits\n        └───motions\n            └─── guoh3dfeats #processed pose data\n            └─── pose_data\n```      \n  \n## Training\nModify the config file in configs folder. The default is set to train the model with DistilBERT.\nChange the line before the main() function accordingly, then train:\n\n```bash\npython train.py\n```\n\nWe prepared four variations of our method based on DistilBert: Original TMR in ''event \u003e event'' (train_bert_orig.yaml) and ''orig \u003e event'' (train_bert_orig_o2e.yaml), our version in ''event \u003e event'' (train_bert_neg.yaml) and ''orig \u003e event'' (train_bert_neg_o2e.yaml). The pretrained weights can be downloaded from [here](https://huggingface.co/line-corporation/ChronAccRet/). The trained models should be placed in ```output``` directory. Please use the config file as guideline to train other variants. To tune the language model, set the `train_text_encoder` option to True.\n\n## Evaluation\nChange the line before the main() function accordingly, then evaluate using conventional metrics:\n```bash\npython evaluate.py\n```\n\nFor CAR accuracy:\n```bash\npython retrieval_shuffle.py\n```\n\nFor motion-to-text retrieval with the shuffled texts:\n```bash\npython evaluate_negative.py\n```\n\n## Acknowledgement\nSome parts of our code are based on [TMR](https://github.com/Mathux/TMR).\n\n\n## Citation\n```bibtex\n@InProceedings{Fujiwara_2024_ECCV, \n\tauthor = {Kent Fujiwara and Mikihiro Tanaka and Qing Yu}, \n\ttitle = {Chronologically Accurate Retrieval for Temporal Grounding of Motion-Language Models}, \n\tbooktitle = {Proc. of the European Conf. on Computer Vision (ECCV)}, \n\tyear = {2024}, \n}\n```\n\n\n## License\n[Apache License 2.0](LICENSE)\n\nAdditionally, this repository contains third-party software. Refer [NOTICE.txt](NOTICE.txt) for more details and follow the terms and conditions of their use.","funding_links":[],"categories":["Uncategorized"],"sub_categories":["Uncategorized"],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fline%2FChronAccRet","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fline%2FChronAccRet","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fline%2FChronAccRet/lists"}