This is an official implement for "HSTforU: Anomaly Detection in Aerial and Ground-based Videos with Hierarchical Spatio-Temporal Transformer for U-net"

A deep learning project to predict and generate future video frames using models like ConvLSTM, PredRNN, and Transformers, leveraging the UCF101 dataset. The repository includes preprocessing, model training, video generation, an interactive UI, and evaluation metrics to compare model performance in video synthesis and temporal prediction tasks.

This is a literature review and code implementation for the video prediction task