Variational Feature Pyramid Networks
Abstract
Recent architectures for object detection adopt a Feature Pyramid Network as a backbone for deep feature extraction. Many works focus on the design of pyramid networks which produce richer feature representations. In this work, we opt to learn a dataset-specific architecture for Feature Pyramid Networks. With the proposed method, the network fuses features at multiple scales, it is efficient in terms of parameters and operations, and yields better results across a variety of tasks and datasets. Starting by a complex network, we adopt Variational Inference to prune redundant connections. Our model, integrated with standard detectors, outperforms the state-of-the-art feature fusion networks.
Cite
Text
Dimitrakopoulos et al. "Variational Feature Pyramid Networks." International Conference on Machine Learning, 2022.Markdown
[Dimitrakopoulos et al. "Variational Feature Pyramid Networks." International Conference on Machine Learning, 2022.](https://mlanthology.org/icml/2022/dimitrakopoulos2022icml-variational/)BibTeX
@inproceedings{dimitrakopoulos2022icml-variational,
title = {{Variational Feature Pyramid Networks}},
author = {Dimitrakopoulos, Panagiotis and Sfikas, Giorgos and Nikou, Christophoros},
booktitle = {International Conference on Machine Learning},
year = {2022},
pages = {5142-5152},
volume = {162},
url = {https://mlanthology.org/icml/2022/dimitrakopoulos2022icml-variational/}
}