利用增强两步迁移学习的多视图融合方法改进内窥镜结石自动识别
图像与视频处理
2023-08-23 v2 计算机视觉与模式识别
机器学习
摘要
本文提出一种深度学习方法,用于提取并融合从不同视角获取的图像信息,以产生更具判别力的对象特征,用于识别内窥镜图像中肾结石的类型。该模型进一步通过两步迁移学习方法及注意力块进行改进,以细化所学特征图。深度特征融合策略在肾结石分类准确率上比较单一视图提取骨干模型提升了逾6%。
引用
@article{arxiv.2304.03193,
title = {Improving automatic endoscopic stone recognition using a multi-view fusion approach enhanced with two-step transfer learning},
author = {Francisco Lopez-Tiro and Elias Villalvazo-Avila and Juan Pablo Betancur-Rengifo and Ivan Reyes-Amezcua and Jacques Hubert and Gilberto Ochoa-Ruiz and Christian Daul},
journal= {arXiv preprint arXiv:2304.03193},
year = {2023}
}
备注
This paper has been accepted at the LatinX in Computer Vision (LXCV) Research workshop at ICCV 2023 (Paris, France)