Attentional Feature Fusion

• Published in: Proceedings / IEEE Workshop on Applications of Computer Vision pp. 3559 - 3568
• Main Authors: Dai, Yimian, Gieseke, Fabian, Oehmcke, Stefan, Wu, Yiquan, Barnard, Kobus
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.01.2021
• Subjects: Computational modeling; Computer vision; Conferences; Fuses; Network architecture; Neural networks; Semantics

Feature fusion, the combination of features from different layers or branches, is an omnipresent part of modern network architectures. It is often implemented via simple operations, such as summation or concatenation, but this might not be the best choice. In this work, we propose a uniform and gene...