Shypd.aiThe largest open index of AI toolsSocial-STGCNN is an open-source Data & Analytics tool that provides a Social Spatio-Temporal Graph Convolutional Neural Network for human trajectory prediction. It offers significant improvements in prediction accuracy and inference speed with fewer parameters.
Trending
About
Social-STGCNN is an open-source implementation of a Social Spatio-Temporal Graph Convolutional Neural Network (Social-STGCNN) designed for human trajectory prediction. This model frames the prediction problem as a spatio-temporal graph, offering substantial improvements over previous methods. It boasts a 20% improvement in Final Displacement Error (FDE) and an improvement in Average Displacement Error (ADE) with 8.5 times fewer parameters and up to 48 times faster inference speed. The model is also data-efficient, achieving state-of-the-art ADE metrics with only 20% of the training data. It includes a kernel function to embed social interactions between pedestrians within the adjacency matrix, and its inference speed is 0.002s/frame (500Hz) using only 7.6K parameters. The architecture comprises two main blocks: an ST-GCNN for spatio-temporal graph embedding and a TXP-CNN for future trajectory extrapolation.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
