OV-SKTGCNN datasets(ETH UVY)

Dataset Repository for: Enhanced Pedestrian Trajectory Prediction via Overlapping Field-of-View Domains and Integrated Kolmogorov-Arnold Networks (OV-SKTGCNN) This repository serves as a documentation hub for the trajectory datasets used in the OV-SKTGCNN model. Note: This repository does NOT contain any code or model implementations - it solely provides dataset provenance information. Dataset Provenance The trajectory data used in our study follows the exact implementation from Social-STGCNN [17], processed from two public pedestrian video sources: Original Video Sources ETH Pedestrian Dataset Original videos published by Pellegrini et al. [27] Access raw videos at: https://data.vision.ee.ethz.ch/cvl/aem/ewap_dataset_full.tgz UCY Crowd Dataset Original videos published by Lerner et al. [28] Access raw videos at: https://graphics.cs.ucy.ac.cy/research/downloads/crowd-data Processed Trajectory Data The structured trajectory files containing: Agent positions in world coordinates Velocity measurements Timestamp information Are available through the Social-STGCNN GitHub repository: https://github.com/abduallahmohamed/Social-STGCNN The data schema and preprocessing methodology are fully described in the original Social-STGCNN paper [17]. Data Compatibility Verification Our OV-SKTGCNN implementation maintains strict compatibility with the original data format: No modifications to data structures Preserved original sampling rates (2.5 Hz) Consistent coordinate normalization Identical train/test splits This ensures direct comparability with baseline models using the same dataset configuration. Usage Notes This repository is provided for: Transparent documentation of data provenance Easy reference for replication studies Comparison with our OV-SKTGCNN results All dataset copyright remains with the original authors. Please observe the license terms from both the original video sources and the Social-STGCNN repository when using this data. Citation When using this data configuration, please cite both the original datasets and the Social-STGCNN processing methodology: @INPROCEEDINGS{5459260, author={Pellegrini, S. and Ess, A. and Schindler, K. and van Gool, L.}, booktitle={2009 IEEE 12th International Conference on Computer Vision}, title={You'll never walk alone: Modeling social behavior for multi-target tracking}, year={2009}, volume={}, number={}, pages={261-268}, keywords={Predictive models;Vehicle dynamics;Layout;Humans;Computer vision;Trajectory;Cameras;Legged locomotion;Path planning;Computer science}, doi={10.1109/ICCV.2009.5459260}}@inproceedings{lerner2007crowds, title={Crowds by example}, author={Lerner, Alon and Chrysanthou, Yiorgos and Lischinski, Dani}, booktitle={Computer graphics forum}, volume={26}, number={3}, pages={655--664}, year={2007}, organization={Wiley Online Library}}@article{almomen2020social, title={Social-STGCNN: A Social Spatio-Temporal Graph Convolutional Neural Network for Human Trajectory Prediction}, author={Almomen, Abduallah and Kim, Kyungtae and How, Jonathan}, journal={arXiv preprint arXiv:2002.11927}, year={2020}}