SwinLSTM:Improving Spatiotemporal Prediction Accuracy using Swin Transformer and LSTM
This work addresses spatiotemporal prediction tasks for researchers and practitioners, offering an incremental advancement by combining existing transformer and LSTM components to improve accuracy.
The paper tackles the problem of spatiotemporal prediction by proposing SwinLSTM, a new recurrent cell that integrates Swin Transformer blocks with a simplified LSTM to capture global spatial dependencies, and it outperforms state-of-the-art methods on datasets like Moving MNIST and Human3.6m, showing significant accuracy improvements over ConvLSTM.
Integrating CNNs and RNNs to capture spatiotemporal dependencies is a prevalent strategy for spatiotemporal prediction tasks. However, the property of CNNs to learn local spatial information decreases their efficiency in capturing spatiotemporal dependencies, thereby limiting their prediction accuracy. In this paper, we propose a new recurrent cell, SwinLSTM, which integrates Swin Transformer blocks and the simplified LSTM, an extension that replaces the convolutional structure in ConvLSTM with the self-attention mechanism. Furthermore, we construct a network with SwinLSTM cell as the core for spatiotemporal prediction. Without using unique tricks, SwinLSTM outperforms state-of-the-art methods on Moving MNIST, Human3.6m, TaxiBJ, and KTH datasets. In particular, it exhibits a significant improvement in prediction accuracy compared to ConvLSTM. Our competitive experimental results demonstrate that learning global spatial dependencies is more advantageous for models to capture spatiotemporal dependencies. We hope that SwinLSTM can serve as a solid baseline to promote the advancement of spatiotemporal prediction accuracy. The codes are publicly available at https://github.com/SongTang-x/SwinLSTM.