linhaoqi027
linhaoqi027
那如果我把decoder直接换成8倍或16倍上采样,这个操作是否有意义。另外请问grad-cam是针对于分类问题提出的嘛?
谢谢您
Towards Interpretable Semantic Segmentation via Gradient-weighted Class Activation Mapping这篇论文有讲到怎么变化就可以实现语义分割可视化。请问大佬这个要复现难吗?
请问这里的index指的是onehot编码下对应类别的index嘛
> 您好, > 我在下载105_checkpoint.pth.tar文件时遇到问题。我尝试在https://pan.baidu.com/上创建一个新帐户,但他们不接受海外手机号码帐户。 > 我要求您将文件上传到另一个平台或保管箱或驱动器。每个人都很容易从那里下载文件。 https://drive.google.com/file/d/1MXPr6WDdlj6ZcqjJClmLUPkE6kCGfuXL/view?usp=sharing
Did you use my code? If yes,it may caused by NMS. Beacuse NMS is not include in openvino.I convert the model output to torch then using NMS code in yolov5...
sorry,I haven't do it. It is a competition using this code. So I upload it in this project.