深度学习论文: RailYolact -- A Yolact Focused on edge for Real-Time Rail Segmentation

深度学习论文: RailYolact – A Yolact Focused on edge for Real-Time Rail Segmentation
RailYolact – A Yolact Focused on edge for Real-Time Rail Segmentation
PDF:https://arxiv.org/pdf/2410.09612
PyTorch代码: https://github.com/shanglianlm0525/CvPytorch
PyTorch代码: https://github.com/shanglianlm0525/PyTorch-Networks

1 概述

为了解决模型预测的铁轨掩码边缘粗糙的问题，本文将边缘算子提取的边缘信息融入原始 Yolact 的损失函数中，以强调模型对铁轨边缘的关注。此外，本文应用盒式滤波器对线性插值导致的真实标签掩码边缘锯齿进行平滑处理。