Das Paper mit dem Titel "Segformer++: Efficient Token-Merging Strategies for High-Resolution Semantic Segmentation" von Daniel Kienzle, Marco Kantonis, Robin Schön und Rainer Lienhart wurde auf der IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) 2024 akzeptiert. Das Paper beschreibt eine neue Methode, um die Effizienz von Transformermodellen zu steigern. Die beschriebenen Methoden ermöglichen den Einsatz von rechenintensiven Transformermodellen für hochauflösender Bilder.

 

Weitere Informationen zu diesem Paper sind unter  https://kiedani.github.io/MIPR2024/ zu finden.

Utilizing transformer architectures for semantic segmentation of high-resolution images is hindered by the attention's quadratic computational complexity in the number of tokens. A solution to this challenge involves decreasing the number of tokens through token merging, which has exhibited remarkable enhancements in inference speed, training efficiency, and memory utilization for image classification tasks. In this paper, we explore various token merging strategies within the framework of the Segformer architecture and perform experiments on multiple semantic segmentation and human pose estimation datasets. Notably, without model re-training, we, for example, achieve an inference acceleration of 61% on the Cityscapes dataset while maintaining the mIoU performance. Consequently, this paper facilitates the deployment of transformer-based architectures on resource-constrained devices and in real-time applications.