This is the second version of the Google Landmarks dataset (GLDv2), which contains images annotated with labels representing human-made and natural landmarks. The dataset can be used for landmark recognition and retrieval experiments. This version of the dataset contains approximately 5 million images, split into 3 sets of images: train, index and test. The dataset was presented in our CVPR'20 paper. In this repository, we present download links for all dataset files and relevant code for metric computation. This dataset was associated to two Kaggle challenges, on landmark recognition and landmark retrieval. Results were discussed as part of a CVPR'19 workshop. In this repository, we also provide scores for the top 10 teams in the challenges, based on the latest ground-truth version. Please visit the challenge and workshop webpages for more details on the data, tasks and technical solutions from top teams.

The Google Landmarks dataset contains 1,060,709 images from 12,894 landmarks, and 111,036 additional query images. The images in the dataset are captured at various locations in the world, and each image is associated with a GPS coordinate. This dataset is used to train and evaluate large-scale image retrieval models.

Dataset Card for Dataset Name

  Dataset Summary

This dataset is a subset of Kaggle's Google Landmark Recognition 2021 competition with only the categories with more than 500 images. https://www.kaggle.com/competitions/landmark-recognition-2021/data The dataset consists of a total of 45579 224x224 color images in 51 categories.

  Languages

English

  Dataset Structure





  Data Fields

landmark_id: Int - Numeric identifier of the category category :… See the full description on the dataset page: https://huggingface.co/datasets/pemujo/GLDv2_Top_51_Categories.

We filter and match the landmarks in the Google Landmarks dataset with their OpenStreetMap polygons and filter for those located in the United States, resulting in 602 landmarks. Then, we obtain the latest high-resolution aerial images of the obtained polygons through the National Agriculture Imagery Program (NAIP) of the United States Department of Agriculture (USDA). Finally, we construct multiple-choice questions about the name of the landmark with incorrect answers from other landmarks in the same category.

Dataset

This dataset was created by Hidehisa Arai

Released under CC0: Public Domain

Contents

Dataset

This dataset was created by Hidehisa Arai

Released under CC0: Public Domain

Contents

Dataset

This dataset was created by kwang

Contents

Context

It took very long time/weeks, to make this dataset, giving me an extensive data engineering capabilities. Used both GitHub and GCP as storage and both kaggle and colab to prepare this dataset. It would have been more useful to everyone, had i done this much earlier.

Content

All images from original set are included. To reduce the dataset size, all images have been resized to a minimum dimension of (224320) using tensorflow resize API.

Acknowledgements

Extensively used stackoverflow to find best solutions for many data engineering tasks and thanks for all those who have solved those issues earlier.

Inspiration

Original dataset size 99GB cant be used in colab to train the custom model.

Dataset

This dataset was created by Mark Wijkhuizen

Contents

This dataset includes buildings and districts in Chicago which are listed on the National Register of Historic Places (NRHP) or designated as National Historic Landmarks (NHL). The NRHP is the official list of the Nation's historic places worthy of preservation; NHLs are nationally significant historic places designated by the Secretary of the Interior because they possess exceptional value or quality in illustrating or interpreting the heritage of the United States. The NRHP and NHL programs are federally-established and are administered by the National Park Service (www.nps/gov/nr) and the Illinois Historic Preservation Agency (IHPA, www.illinoishistory.gov/). This dataset is provided by the City of Chicago based on NRHP and NHL nominations provided by IHPA. To view or use this KMZ file, compression software, such as 7-Zip, and special GIS software, such as Google Earth, are required. To download this file, right-click the "Download" link above and choose "Save link as." Time Period: Data is current as of June 2012. Update Frequency: Data is updated as needed.

Individual landmarks in Chicago. The data can be viewed on the Chicago Data Portal with a web browser. However, to view or use the files outside of a web browser, you will need to use compression software and special GIS software, such as ESRI ArcGIS (shapefile) or Google Earth (KML or KMZ).

Analysis of ‘Individual Landmarks’ provided by Analyst-2 (analyst-2.ai), based on source dataset retrieved from https://catalog.data.gov/dataset/496ed9f5-90dd-413a-be8e-3d1aaa5d2646 on 26 January 2022.

--- Dataset description provided by original source is as follows ---

Individual landmarks in Chicago. The data can be viewed on the Chicago Data Portal with a web browser. However, to view or use the files outside of a web browser, you will need to use compression software and special GIS software, such as ESRI ArcGIS (shapefile) or Google Earth (KML or KMZ).

--- Original source retains full ownership of the source dataset ---

Dataset

This dataset was created by Nawid Sayed

Contents

description: KML file for landmark districts in Chicago. To view or use these files, special GIS software such Google Earth is required.; abstract: KML file for landmark districts in Chicago. To view or use these files, special GIS software such Google Earth is required.

Dataset Card for "hagrid-mediapipe-hands"

This dataset is designed to train a ControlNet with human hands. It includes hand landmarks detected by MediaPipe(for more information refer to: https://developers.google.com/mediapipe/solutions/vision/hand_landmarker). The source image data is from HaGRID dataset and we use a modified version from Kaggle(https://www.kaggle.com/datasets/innominate817/hagrid-classification-512p) to build this dataset. There are 507050 data samples in total… See the full description on the dataset page: https://huggingface.co/datasets/Vincent-luo/hagrid-mediapipe-hands.

Individual Chicago Landmarks designated by City Council upon recommendation of the Commission on Chicago Landmarks. To view or use these files, special GIS software such as Google Earth is required. To download, right-click the "Download" link above and choose "Save link as."

The Chicago Historic Resources Survey (CHRS), completed in 1995, was a decade-long research effort by the City of Chicago to analyze the historic and architectural importance of all buildings, objects, structures, and sites constructed in the city prior to 1940. During 12 years of field work and follow-up research that started in 1983, CHRS surveyors identified approximately 9,900 properties which were considered to have some historic or architectural importance. Please note that this CHRS dataset is limited and does not include the entire survey:

1. A color-coded ranking system was used to identify historic and architectural significance relative to age, degree of external physical integrity, and level of possible significance. This dataset only includes buildings identified with the two highest color codes: "Red" and "Orange." Buildings and structures coded "Red" or "Orange" (unless designated as a Chicago Landmark or located within a Chicago Landmark District) are subject to the City of Chicago’s Demolition-Delay Ordinance (link to: http://www.cityofchicago.org/city/en/depts/dcd/supp_info/demolition_delay.html), adopted by City Council in 2003.

2. Only buildings are included in this dataset; structures and objects such as bridges, park structures, monuments and mausoleums, generally are not represented. Likewise, garages, coach houses, and other secondary structures associated with a building may not be consistently depicted or color-coded. If an “Orange”- or “Red”-rated building was demolished after 2008, it may still appear in the map. The CHRS occasionally rated only part of a building or part of a group of joined buildings as “Orange” or “Red;” however the entire building or group of joined buildings may be incorrectly identified as “Orange” or “Red.”

Additional information about the CHRS is available at www.cityofchicago.org/Landmarks/ or by contacting the Historic Preservation Division at (312) 744-3200.

To view or use this KMZ file, compression software, such as 7-Zip, and special GIS software, such as Google Earth, are required. To download this file, right-click the "Download" link above and choose "Save link as."

Image pose detectors in which the pose if defined by anatomical landmarks are rare and scarcely available, which impedes progress in methods of markerless motion capture and analysis applied to biomechanics. Temporal 3D scanning (or 4D scanning) systems allow human bodies in motion to be obtained with high precision, as well as providing a realistic 3D avatar of the scanned person. These two aspects can be used to obtain the pose by different methods. A first method is the location of anatomical landmarks on the mesh surface in a direct way. The other method consists in obtaining virtual images of the mesh from different points of view from which a neural network can estimate the location of body markers.

This dataset associates 2D and 3D human pose keypoints estimated from images with MediaPipe (https://developers.google.com/mediapipe) with the location of their corresponding 3D anatomical landmarks.

It consists of 567 movement sequences of 71 participants in A-Pose and performing 7 movements (walking, running, squatting, and four types of jump) who were scanned with Move4D (https://www.move4d.net/) to build a collection of 3D human moving meshes with texture and with anatomical correspondence (a total amount of 50,952 poses). From each mesh of that collection, the 3D locations of 53 anatomical landmarks were obtained and 48 images were created using virtual cameras with different perspectives. 2D pose keypoints from those images were obtained using MediaPipe's Pose landmarker model and their corresponding 3D keypoints were calculated by linear triangulation.

There is one folder per participant which contains two Track Row Column (TRC) files and one JSON file per movement sequence. One TRC file is used to store the 3D keypoints triangulated and the other contains the 3D anatomical landmarks. The JSON file stores 2D keypoints and the calibration parameters of the virtual cameras. The anthropometric characteristics of the participants (height, weight, age and sex) are annotated in a single CSV file.

The files are named following the next scheme [CODE]_[MOVEMENT]_[DATA].[EXT], in which: [CODE]: is the participant code. The last character (F/M) makes reference to the sex ("F" female, "M" male). [MOVEMENT]: Name of the movement (A-POSE / F-JUMP / GAIT / J-JACKS / JUMP / RUNNING / SQUATS / T-JUMP). [DATA]: Reference data (AL = anatomical landmarks / KP2D = keypoints in pixels and cameras calibration / KP3D = 3D keypoints obtained by triangulating the 2D keypoints). [EXT]: file extension (TRC / JSON).

The JSON files contains general sequence data ("subject", "movement", "fps") and a list of annotations ("frame", "camera", "keypoint_scores", "proj_matrix", "proj_matrix_rows", "proj_matrix_cols" ). "keypoint_scores" is vector of size 99 (u1, v1, score1, … u33, v33, score33) where (ui,vi) is the 2D keypoint location and score(i) is its associated score.

The AVI file shows an example of the 3D landmarks contained in both TRC for TDB_001_F.

Dataset

This dataset was created by Mark Wijkhuizen

Contents

Forest product information and landmarks