[MICCAI, 2024] (Oral, RU @ MedIA Best Paper Award) Official implementation of the Brain Latent Progression (BrLP) method from "Enhancing Spatiotemporal Disease Progression Models via Latent Diffusion and Prior Knowledge"
[ECCV 2024] ViewFormer: Exploring Spatiotemporal Modeling for Multi-View 3D Occupancy Perception via View-Guided Transformers
BSTVC-R is an R package for spatiotemporal heterogeneous analysis within a unified full-map framework. It supports the analysis of locally varying influencing factors, identification of key global drivers, and dynamic prediction, providing a programmable and reproducible workflow for spatiotemporal interpretable modeling.
Model evolution for short-term maritime traffic forecasting from AIS grid sequences.
Framework for benchmarking spatiotemporal models for global ionosphere forecasting
Python based Jupyter notebook for reconstruction of XCO2 using a Transformer-based model for emission monitoring, integrating OCO-2 data with land-atmosphere variables to provide environmental context.
An implementation of the Semiparametric Spatiotemporal Model with Mixed Frequencies
ST-GraphAD Source Code : Spatio-Temporal Graph Neural Network for network intrusion detection on the UNSW-NB15 dataset. The system converts network flows into dynamic attributed graphs and uses a 3-layer Graph Convolutional Network (GCN) combined with a 2-layer, 4-head temporal attention mechanism to detect anomalies.
A patch-augmented HybridCNNLSTM pipeline for wetland methane emission upscaling using X-MethaneWet, TEM-MDM simulation data, and FLUXNET-CH4 observations.
Hybrid CNN-ViT spatiotemporal driver drowsiness detection pipeline using MobileViT v2 frame features, temporal sequence modeling, and video-level soft voting.
[AAAI] A spatiotemporal embedding framework for geographical entities
Spatiotemporal engression for probabilistic forecasting on low-frequency datasets.
SAFE: Spatial-temporal Adaptation in Federated Environments for CAV Accident Anticipation under Heterogeneity
in progress research paper
End-to-end NYC taxi demand forecasting pipeline featuring EDA, feature engineering, spatial clustering, MLflow experiment tracking, and LightGBM/CatBoost/XGBoost model development on 33M+ records.
Reproducible time-conditioned U-Net surrogate for two-dimensional Richards-equation soil infiltration, with a qualified numerical reference, a 208-scenario corpus, locked evaluation, and portable inference.
Graph-, flow-, and physics-aware XGBoost for flood forecasting
Cryptocurrency closing price forecasting using a spatio-temporal graph neural network (T-MTGNN).
AI/ML epidemiology portfolio for infectious disease outbreak forecasting, surveillance modeling, and reproducible public health analytics.
To associate your repository with the spatiotemporal-modeling topic, visit your repo's landing page and select "manage topics."