{"id":18488218,"url":"https://github.com/choyingw/gais-net","last_synced_at":"2025-08-22T13:13:04.437Z","repository":{"id":112372181,"uuid":"328938460","full_name":"choyingw/GAIS-Net","owner":"choyingw","description":"CVPR 2020 Workshop on Scalability in Autonomous Driving: GAIS-Net: Geometry-Aware Instance Segmentation with Disparity Maps","archived":false,"fork":false,"pushed_at":"2024-03-28T04:53:59.000Z","size":973,"stargazers_count":8,"open_issues_count":4,"forks_count":0,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-03-23T17:51:31.051Z","etag":null,"topics":["3d","autonomous-vehicles","computer-vision","cvpr2020","deep-neural-networks","geometry-processing","instance-segmentation","mask-rcnn","multimodal-deep-learning","scalability","stereo-vision"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/choyingw.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,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2021-01-12T09:40:44.000Z","updated_at":"2024-04-28T03:33:57.000Z","dependencies_parsed_at":"2024-11-06T12:54:34.616Z","dependency_job_id":"5a67d611-0de6-40bf-aef5-6f3f1a47c2e8","html_url":"https://github.com/choyingw/GAIS-Net","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/choyingw%2FGAIS-Net","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/choyingw%2FGAIS-Net/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/choyingw%2FGAIS-Net/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/choyingw%2FGAIS-Net/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/choyingw","download_url":"https://codeload.github.com/choyingw/GAIS-Net/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247923408,"owners_count":21018988,"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":["3d","autonomous-vehicles","computer-vision","cvpr2020","deep-neural-networks","geometry-processing","instance-segmentation","mask-rcnn","multimodal-deep-learning","scalability","stereo-vision"],"created_at":"2024-11-06T12:51:25.663Z","updated_at":"2025-04-08T20:32:39.411Z","avatar_url":"https://github.com/choyingw.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# GAIS-Net\n\nGAIS-Net: Geometry-Aware Instance Segmentation with Disparity Maps\n\nCho-Ying Wu, Xiaoyan Hu, Michael Happold, Qiangeng Xu, Ulrich Neumann, CVPR Workshop on Scalability in Autonomous Driving, 2020.\n\n**Advantage:** \n\n**:+1: The first work to directly regress instances from depth maps to advance the multi-modal learning for outdoor scenarios.**\n\n**:+1: High performance and highly modulized. The codes are based on the mask-rcnn benchmark framework.**\n\nPlease visit out [Project site](https://choyingw.github.io/works/GAIS-Net/index.html) for paper and intorduction.\n\nThis project is developed upon [Mask-RCNN](https://github.com/facebookresearch/maskrcnn-benchmark) and is done during an internship at [Argo AI](https://www.argo.ai/)\n\n\u003cimg src='teaser.png'\u003e\n\nFeature: Resolve overlapping areas between instances by introducing geometry information\n\n\n# Installation\n\nCheck [INSTALL.md](INSTALL.md) for installation instructions.\n\nNote that cocoapi, cityscapesScripts, and apex are needed for the evaluation.\n\n\n# Implementation\n\nThe geometry-aware fusion module is implemented under [roi_heads.py](maskrcnn_benchmark/modeling/roi_heads/roi_heads.py) [mask_head.py](maskrcnn_benchmark/modeling/roi_heads/mask_head/mask_head.py) [maskiou_head.py](maskrcnn_benchmark/modeling/roi_heads/maskiou_head/maskiou_head.py)\n\nCheck these files and related files for the features we implement.\n\n# Data\n\n1. The pre-generated stereo disparity maps by PSMNet for Cityscapes could be downloaded [here](https://drive.google.com/file/d/1yeDkcrl9t3QO0K2NQxfjxcHhtJR4hwYY/view?usp=sharing). Please first create folders \"datasets/cityscapes\" at the root of the repo and extract the zip file under datasets/cityscapes/\n\n2. Go to [Cityscapes](https://www.cityscapes-dataset.com/) to download images of train/val set. Put the images under \"datasets/cityscapes/train/image_2\" and also \"val\". Download the annotations and put under datasets/cityscapes/annotations/\n\n\n# Evaluation\n\nNUM_GPUS=4 \npython -m torch.distributed.launch --nproc_per_node=$NUM_GPUS ./tools/test_net.py --config-file \"configs/cityscapes_v4.yaml\" TEST.IMS_PER_BATCH 4\n\n# Pretrain model\n\nThe pretrained weights could be downloaded [here](https://drive.google.com/file/d/1ZETFaG_xxw0NsX8S9Tj10Rp-XwQNsenf/view?usp=sharing) put the .pth file under \"ckpt/\"\n\n# Results\n\n\u003cimg src='Qualitative.png'\u003e\n\u003cimg src='results.png'\u003e\n\n\n# Citations\n\nIf you find the work useful, please condisder to cite:\n\n\t@inproceedings{wu2020Cvprw,\n\ttitle={Geometry-Aware Instance Segmentation with Disparity Maps},\n\tauthor={Wu, Cho-Ying and Hu, Xiaoyan and Happold, Michael and Xu, Qiangeng and Neumann, Ulrich},\n\tbooktitle={CVPR Workshop on Scability in Autonomous Driving},\n\tyear={2020}\n\t}\n\n# Thanks to the Third Parties\n\nThank [Mask-RCNN](https://github.com/facebookresearch/maskrcnn-benchmark) and [Mask Scoring RCNN](https://github.com/zjhuang22/maskscoring_rcnn)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fchoyingw%2Fgais-net","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fchoyingw%2Fgais-net","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fchoyingw%2Fgais-net/lists"}