The paper titled "Segformer++: Efficient Token-Merging Strategies for High-Resolution Semantic Segmentation" by Daniel Kienzle, Marco Kantonis, Robin Schön, and Rainer Lienhart has been accepted at the IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) 2024. The paper describes a new method to enhance the efficiency of transformer models. This enables the application of computationally intensive transformer models to high-resolution images.

 

Further information about this paper can be found at https://kiedani.github.io/MIPR2024/.

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.