{"id":24396827,"url":"https://github.com/ai4ce/nyu-vpr","last_synced_at":"2026-03-10T04:34:09.433Z","repository":{"id":67192850,"uuid":"389234360","full_name":"ai4ce/NYU-VPR","owner":"ai4ce","description":"[IROS2021] NYU-VPR: Long-Term Visual Place Recognition Benchmark with View Direction and Data Anonymization Influences","archived":false,"fork":false,"pushed_at":"2025-03-14T14:43:13.000Z","size":75540,"stargazers_count":32,"open_issues_count":0,"forks_count":3,"subscribers_count":5,"default_branch":"main","last_synced_at":"2025-06-04T04:15:02.324Z","etag":null,"topics":["benchmark","dataset","dbow","netvlad","visual-place-recognition","vlad","vpr"],"latest_commit_sha":null,"homepage":"https://ai4ce.github.io/NYU-VPR","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ai4ce.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"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,"zenodo":null}},"created_at":"2021-07-25T01:38:34.000Z","updated_at":"2025-04-13T13:18:56.000Z","dependencies_parsed_at":null,"dependency_job_id":"5dae0ad2-0aa6-4cc1-adb0-cbeeb130b821","html_url":"https://github.com/ai4ce/NYU-VPR","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/ai4ce/NYU-VPR","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ai4ce%2FNYU-VPR","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ai4ce%2FNYU-VPR/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ai4ce%2FNYU-VPR/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ai4ce%2FNYU-VPR/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ai4ce","download_url":"https://codeload.github.com/ai4ce/NYU-VPR/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ai4ce%2FNYU-VPR/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30324452,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-10T01:36:58.598Z","status":"online","status_checked_at":"2026-03-10T02:00:06.579Z","response_time":106,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"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":["benchmark","dataset","dbow","netvlad","visual-place-recognition","vlad","vpr"],"created_at":"2025-01-19T21:58:37.635Z","updated_at":"2026-03-10T04:34:09.423Z","avatar_url":"https://github.com/ai4ce.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# NYU-VPR\n\nThis repository provides the experiment code for the paper [Long-Term Visual Place Recognition Benchmark with View Direction and Data Anonymization Influences](https://arxiv.org/abs/2110.09004).\n\nHere is a graphical user interface (GUI) for using VPR methods on custom datasets: https://github.com/ai4ce/VPR-GUI-Tool\n\n## Requirements\n\nTo install requirements:\n\n```\npip install -r requirements.txt\n```\n\n\n\n## Data Processing\n\n**1. Image Anonymization**\n\nTo install mseg-api:\n\n```\ncd segmentation\ncd mseg-api\npip install -e .\n```\n\nMake sure that you can run `python -c \"import mseg\"` in python.\n\nTo install mseg-semantic:\n\n```\ncd segmentation\ncd apex\npip install -v --no-cache-dir --global-option=\"--cpp_ext\" --global-option=\"--cuda_ext\" ./\n\ncd ../mseg-semantic\npip install -e .\n```\n\nMake sure that you can run `python -c \"import mseg_semantic\"` in python.\n\nFinally:\n\n```\ninput_file=/path/to/my/directory\nmodel_name=mseg-3m\nmodel_path=mseg_semantic/mseg-3m.pth\nconfig=mseg_semantic/config/test/default_config_360_ms.yaml\npython -u mseg_semantic/tool/universal_demo.py --config=${config} model_name {model_name} model_path ${model_path} input_file ${input_file}\n```\n\n**2. Image Filtration**\n\nInside the `process` folder, use `whiteFilter.py` to filter images with white pixel percentage.\n\n## Methods\n\n**1. VLAD+SURF**\n\nModify `vlad_codebook_generation.py` line 157 - 170 to fit the dataset.\n\n```\ncd test/vlad\npython vlad_codebook_generation.py\npython query_image_closest_image_generation.py\n```\n\n*Notice: the processing may take a few hours.\n\n**2. VLAD+SuperPoint**\n\n```\ncd test/vlad_SP\npython main.py\npython find_closest.py\n```\n\n*Notice: the processing may take a few hours.\n\n**3. NetVLAD**\n\n**4. PoseNet**\n\nCopy the `train_image_paths.txt` and `test_image_paths.txt` to test/posenet.\n\nObtain the latitude and longtitude of training images and convert them to normalized Universal Transverse Mercator (UTM) coordinates.\n\n```\ncd test/posenet\npython getGPS.py\npython mean.py\n```\n\nStart training. This may take several hours. Suggestion: use slurm to run the process.\n\n```\npython train.py --image_path path_to_train_images/ --metadata_path trainNorm.txt\n```\n\nGenerate the input file for testing from test_image_paths.txt.\n\n```\npython gen_test_txt.py\n```\n\nStart testing.\n```\npython single_test.py --image_path path_to_test_images/ --metadata_path test.txt --weights_path models_trainNorm/best_net.pth\n```\n\nThe predicted normalized UTM coordinates of test images is in the image_name.txt. Match the test images with the training images based on their location.\n\n```\npython match.py\n```\n\nThe matching result is in the match.txt.\n\n**5. DBoW**\n\nCopy the train_image_paths.txt and test_image_paths.txt to test/DBow3/utils. Copy and paste the content of test_image_paths.txt at the end of train_image_paths.txt and save the text file as total_images_paths.txt.\n\nOpen test/DBow3/utils/demo_general.cpp file. Change the for loop range at line 117 and line 123. Both ranges are the range of lines in total_images_paths.txt. The first for loop range is the range of test images and the second range is the range of training images. To run with multi-thread, you may run the code multiple times with small ranges of test images where the sum of ranges equals to the number of lines in test_image_paths.txt.\n\nCompile and run the code.\n\n```\ncd test/DBow3\ncmake .\ncd utils\nmake\n./demo_general a b\n```\n\nThe result of each test image and its top-5 matched training images is in the output.txt.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fai4ce%2Fnyu-vpr","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fai4ce%2Fnyu-vpr","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fai4ce%2Fnyu-vpr/lists"}