DATASETS – Construction Motion Library (CML)

The developed CML dataset contains 225 types of activities and 146,480 samples; among them, 60 types of activities and 61,275 samples are highly related to construction activities. The developers have tested five widely applied deep learning algorithms were adopted to examine the dataset, and the usability, quality, and sufficiency were reported. The average accuracy of models without tunning can reach 74.62% to 83.92%.

Cite: Tian, Y., Li, H., Cui, H., and Chen, J.,  “Construction motion data library: an integrated motion dataset for on-site activity recognition.”, Sci Data, 9, 726 (2022). https://doi.org/10.1038/s41597-022-01841-1

Link

Dataset URL

Figshare – “Construction motion data library: an integrated motion dataset for on-site activity recognition. “,  https://doi.org/10.6084/m9.figshare.20480787.v3

Relevant Packages

To build CML, the researchers utilized Mathwork Matlab 2020a to parse and export the ASF/AMC and BVH files.  The developed code can be accessed with https://github.com/YUANYUAN2222/GIT_json_to_BVH. Meanwhile, the code could be used to retag and process different datasets (i.e., Resampling and Skeletal structure alignment) also can be found via https://github.com/YUANYUAN2222/Integrated-public-3D-skeleton-form-CML-library  , this allow users to prepare the own customizable datasets.

*Note: The developed package used following open-source packages

– https://github.com/lawrennd/mocap

– https://github.com/HW140701/VideoTo3dPoseAndBvh

DATASETS – Tiled Multi-city Urban Objects Dataset

City-scale building energy simulation provides a significant reference for planning and urban management. However, large-scale building energy simulation is often unfeasible due to the huge amount of computational resources required and the lack of high-precision building models. For such reasons, this study developed a tiled multi-city urban objects dataset and a distributed data ontology. Such a data metric not only transforms the conventional whole-city simulation model into patch-based distributed simulations but also incorporates interactive relationships among objects in cities. The dataset stores urban objects (8,196,003 buildings; 238,736 vegetations; 2,381,6698 streets; 430,364 UrbanTiles; 430,464 UrbanPatches) from thirty major cities in the United States. It also aggregated morphological features for each UrbanTile.

Cite: Ma, R., Fang, D., Chen, J., and Li, X., A tiled multi-city urban objects dataset for city-scale building energy simulation. Sci Data 10, 352 (2023). https://doi.org/10.1038/s41597-023-02261-5

Link

Dataset URL

Figshare – “A tiled multi-city urban objects dataset for city-scale building energy simulation V3. “,  https://doi.org/10.6084/m9.figshare.20799637.v3

Relevant Packages

The shared dataset is prepared based on the default setting of the UrbanPatch container and D-radius. If users want to customizable this dataset with different settings, they can use the shared UrbanPatch generation package (https://github.com/ruirzma/UPTO). There are four files included in the package:

  • “ConPatchForTile.py”: construct UrbanPatch individuals for UrbanTile objects when changing the receptive radius.
  • “ConPatchForBuilding.py”: construct UrbanPatch individuals for Building objects for a given receptive radius.
  • “GenMicroclimate.py”: generate the UrbanTile-scale microclimate.
  • “GenIDF.py”: generate UrbanTile-scale EnergyPlus IDF file.

Note!

All tools, packages, protocols and datasets developed by HBI-lab follow open-source principles. Any usage and publication should cite the source properly.