… for the machine translation, where each transformer layer contains a single (… image transformer for imagecaptioning, where each transformer layer implements multiple sub-transformers…
G Li, L Zhu, P Liu, Y Yang - Proceedings of the IEEE/CVF …, 2019 - openaccess.thecvf.com
… In imagecaptioning, the typical attention mechanisms are … are the dominating architectures in imagecaptioning. However, … In this paper, we investigate a Transformer-based sequence …
… and efficiently incorporate grid features with transformer-based architecture for image captioning. To achieve this target, we propose a S2 Transformer—a simple yet effective approach …
T Xian, Z Li, C Zhang, H Ma - Neural Networks, 2022 - Elsevier
… visual representation, which is essential to improve captioning performance. In GED, we first … Hence, we regard the existing captions generated by the classical imagecaptioning model …
X Yang, Y Liu, X Wang - Proceedings of the 30th ACM International …, 2022 - dl.acm.org
… of imagecaptioning, we propose a novel architecture ReFormer- a RElational transFORMER to … in the image. ReFormer incorporates the objective of scene graph generation with that of …
… Dual-Level Collaborative Transformer In this section, we introduce a novel imagecaptioning model, named Dual-Level Collaborative Transformer, which uses both grid and region …
W Liu, S Chen, L Guo, X Zhu, J Liu - arXiv preprint arXiv:2101.10804, 2021 - arxiv.org
… In this paper, we consider the imagecaptioning task from a new sequence-to-sequence … TransformeR (CPTR) which takes the sequentialized raw images as the input to Transformer. …
… ImageCaptioning is the task of translating an input image into a textual description. As such… In this work, we focus on Transformer-based imagecaptioning models and provide …
Y Zhou, Y Zhang, Z Hu, M Wang - Proceedings of the IEEE …, 2021 - openaccess.thecvf.com
… captioning models. We evaluate SATIC model on the challenging MSCOCO [3] image captioning … We present three examples of generated imagecaptions in Figure 3. From the top …