{"id":13936087,"url":"https://github.com/driving-behavior/DBNet","last_synced_at":"2025-07-19T21:31:37.652Z","repository":{"id":215159215,"uuid":"131108755","full_name":"driving-behavior/DBNet","owner":"driving-behavior","description":"DBNet: A Large-Scale Dataset for Driving Behavior Learning, CVPR 2018","archived":false,"fork":false,"pushed_at":"2019-03-20T08:37:59.000Z","size":393,"stargazers_count":214,"open_issues_count":6,"forks_count":50,"subscribers_count":16,"default_branch":"master","last_synced_at":"2024-11-27T04:30:34.287Z","etag":null,"topics":["autonomous-driving","benchmark","cvpr2018","dbnet","driving-behavior","point-cloud","steering-wheel","vehicle-speed"],"latest_commit_sha":null,"homepage":"http://www.dbehavior.net/","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/driving-behavior.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,"roadmap":null,"authors":null}},"created_at":"2018-04-26T06:16:08.000Z","updated_at":"2024-11-18T05:13:10.000Z","dependencies_parsed_at":"2024-01-06T01:04:54.319Z","dependency_job_id":null,"html_url":"https://github.com/driving-behavior/DBNet","commit_stats":null,"previous_names":["driving-behavior/dbnet"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/driving-behavior/DBNet","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/driving-behavior%2FDBNet","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/driving-behavior%2FDBNet/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/driving-behavior%2FDBNet/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/driving-behavior%2FDBNet/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/driving-behavior","download_url":"https://codeload.github.com/driving-behavior/DBNet/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/driving-behavior%2FDBNet/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":266019657,"owners_count":23864916,"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":["autonomous-driving","benchmark","cvpr2018","dbnet","driving-behavior","point-cloud","steering-wheel","vehicle-speed"],"created_at":"2024-08-07T23:02:22.237Z","updated_at":"2025-07-19T21:31:32.638Z","avatar_url":"https://github.com/driving-behavior.png","language":"Python","funding_links":[],"categories":["Datasets","Python"],"sub_categories":["Sensor and Acuator Interfaces"],"readme":"\u003ca href=\"http://www.dbehavior.net/\"\u003e\u003cimg src=docs/logo.jpeg width=135/\u003e\u003c/a\u003e\n\n![db-prediction](docs/pred.jpg)\n\n[DBNet](http://www.dbehavior.net/) is a __large-scale driving behavior dataset__, which provides large-scale __high-quality point clouds__ scanned by Velodyne lasers, __high-resolution videos__ recorded by dashboard cameras and __standard drivers' behaviors__ (vehicle speed, steering angle) collected by real-time sensors.\n\nExtensive experiments demonstrate that extra depth information helps networks to determine driving policies indeed. We hope it will become useful resources for the autonomous driving research community.\n\n_Created by [Yiping Chen*](https://scholar.google.com/citations?user=e9lv2fUAAAAJ\u0026hl=en), [Jingkang Wang*](https://wangjksjtu.github.io/), [Jonathan Li](https://uwaterloo.ca/mobile-sensing/people-profiles/jonathan-li), [Cewu Lu](http://www.mvig.org/), Zhipeng Luo, HanXue and [Cheng Wang](http://chwang.xmu.edu.cn/). (*equal contribution)_\n\nThe resources of our work are available: [[paper]](http://openaccess.thecvf.com/content_cvpr_2018/papers/Chen_LiDAR-Video_Driving_Dataset_CVPR_2018_paper.pdf), [[code]](https://github.com/driving-behavior/DBNet), [[video]](http://www.dbehavior.net/data/demo.mp4), [[website]](http://www.dbehavior.net/), [[challenge]](http://www.dbehavior.net/task.html), [[prepared data]](https://drive.google.com/file/d/1WxzOrhvMnHCOkh6EFGWltflyPb_UnGqo/view?usp=sharing)\n\n\u003c!--\n## News!\n__DBNet Autonomous Driving Data (prepared \u0026 raw) are released [here](http://www.dbehavior.net/download.aspx)!__\n___We are going to organize DBNet challenges for CVPR/ICCV/ECCV Workshops. The instructions of DBNet-2018 challenge will be open soon. Stay tuned!___\n--\u003e\n\n## Contents\n1. [Introduction](#introduction)\n2. [Requirements](#requirements)\n3. [Quick Start](#quick-start)\n4. [Baseline](#baseline)\n5. [Contributors](#contributors)\n6. [Citation](#citation)\n7. [License](#license)\n\n## Introduction\nThis work is based on our [research paper](http://openaccess.thecvf.com/content_cvpr_2018/html/Chen_LiDAR-Video_Driving_Dataset_CVPR_2018_paper.html), which appears in CVPR 2018. We propose a large-scale dataset for driving behavior learning, namely, DBNet. You can also check our [dataset webpage](http://www.dbehavior.net/) for a deeper introduction.\n\nIn this repository, we release __demo code__ and __partial prepared data__ for training with only images, as well as leveraging feature maps or point clouds. The prepared data are accessible [here](https://drive.google.com/open?id=14RPdVTwBTuCTo0tFeYmL_SyN8fD0g6Hc). (__More demo models and scripts are released soon!__)\n\n## Requirements\n\n* **Tensorflow 1.2.0**\n* Python 2.7\n* CUDA 8.0+ (For GPU)\n* Python Libraries: numpy, scipy and __laspy__\n\nThe code has been tested with Python 2.7, Tensorflow 1.2.0, CUDA 8.0 and cuDNN 5.1 on Ubuntu 14.04. But it may work on more machines (directly or through mini-modification), pull-requests or test report are well welcomed.\n\n## Quick Start\n### Training\nTo train a model to predict vehicle speeds and steering angles:\n\n    python train.py --model nvidia_pn --batch_size 16 --max_epoch 125 --gpu 0\n\nThe names of the models are consistent with our [paper](http://www.dbehavior.net/publications.html).\nLog files and network parameters will be saved to `logs` folder in default.\n\nTo see HELP for the training script:\n\n    python train.py -h\n\nWe can use TensorBoard to view the network architecture and monitor the training progress.\n\n    tensorboard --logdir logs\n\n### Evaluation    \nAfter training, you could evaluate the performance of models using `evaluate.py`. To plot the figures or calculate AUC, you may need to have matplotlib library installed.\n\n    python evaluate.py --model_path logs/nvidia_pn/model.ckpt\n\n### Prediction\nTo get the predictions of test data:\n\n    python predict.py\n\nThe results are saved in `results/results` (every segment) and `results/behavior_pred.txt` (merged) by default.\nTo change the storation location:\n\n    python predict.py --result_dir specified_dir\n\nThe result directory will be created automatically if it doesn't exist.\n\n## Baseline\n\u003ctable style=\"undefined;table-layout: fixed; width: 512px\"\u003e\u003ccolgroup\u003e\u003ccol style=\"width: 68px\"\u003e\u003ccol style=\"width: 106px\"\u003e\u003ccol style=\"width: 66px\"\u003e\u003ccol style=\"width: 88px\"\u003e\u003ccol style=\"width: 54px\"\u003e\u003ccol style=\"width: 46px\"\u003e\u003ccol style=\"width: 38px\"\u003e\u003ccol style=\"width: 46px\"\u003e\u003c/colgroup\u003e\u003ctr\u003e\u003cth\u003eMethod\u003c/th\u003e\u003cth colspan=\"2\"\u003eSetting\u003c/th\u003e\u003cth\u003eAccuracy\u003c/th\u003e\u003cth\u003eAUC\u003c/th\u003e\u003cth\u003eME\u003c/th\u003e\u003cth\u003eAE\u003c/th\u003e\u003cth\u003eAME\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd rowspan=\"2\"\u003envidia-pn\u003c/td\u003e\u003ctd rowspan=\"2\"\u003eVideos + Laser Points\u003c/td\u003e\u003ctd\u003eangle\u003c/td\u003e\u003ctd\u003e70.65% (\u0026lt;5)\u003c/td\u003e\u003ctd\u003e0.7799 \u003c/td\u003e\u003ctd\u003e29.46\u003c/td\u003e\u003ctd\u003e4.23\u003c/td\u003e\u003ctd\u003e20.88\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd\u003espeed\u003c/td\u003e\u003ctd\u003e82.21% (\u0026lt;3)\u003c/td\u003e\u003ctd\u003e0.8701\u003c/td\u003e\u003ctd\u003e18.56\u003c/td\u003e\u003ctd\u003e1.80\u003c/td\u003e\u003ctd\u003e9.68\u003c/td\u003e\u003c/tr\u003e\u003c/table\u003e\n\nThis baseline is run on __dbnet-2018 challenge data__ and only __nvidia\\_pn__ is tested. To measure difficult architectures comprehensively, several metrics are set, including accuracy under different thresholds, area under curve (__AUC__), max error (__ME__), mean error (__AE__) and mean of max errors (__AME__).\n\nThe implementations of these metrics could be found in `evaluate.py`.\n\n## Contributors\nDBNet was developed by [MVIG](http://www.mvig.org/), Shanghai Jiao Tong University* and [SCSC](http://scsc.xmu.edu.cn/) Lab, Xiamen University* (*alphabetical order*).\n\n## Citation\nIf you find our work useful in your research, please consider citing:\n\n\t@InProceedings{DBNet2018,\n\t  author = {Yiping Chen and Jingkang Wang and Jonathan Li and Cewu Lu and Zhipeng Luo and HanXue and Cheng Wang},\n\t  title = {LiDAR-Video Driving Dataset: Learning Driving Policies Effectively},\n\t  booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},\n\t  month = {June},\n\t  year = {2018}\n\t}\n\n## License\nOur code is released under Apache 2.0 License. The copyright of DBNet could be checked [here](http://www.dbehavior.net/contact.html).\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdriving-behavior%2FDBNet","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fdriving-behavior%2FDBNet","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdriving-behavior%2FDBNet/lists"}