Abstract: This research investigates Time-Series Transformer architectures for Electrocardiogram (ECG) heartbeat classification, particularly focusing on their generalization capabilities towards new ...
@article{zhang2025unified, title={Unified Multimodal Understanding and Generation Models: Advances, Challenges, and Opportunities}, author={Zhang, Xinjie and Guo, Jintao and Zhao, Shanshan and Fu, ...
2025.09.26: Release the real-world EDTR detection model and demo examples. The code has also been simplified. 2025.09.01: The code is released. Our code has been tested with PyTorch 2.2.2 and CUDA ...
The proposed Coordinate-Aware Feature Excitation (CAFE) module and Position-Aware Upsampling (Pos-Up) module both adhere to ...
Abstract: Transformers are emerging as a powerful alternative to convolutional neural networks (CNNs) for hyperspectral image (HSI) classification. However, most existing approaches either neglect the ...