cfzd
cfzd
@luweiqing 你确定你用来编译的cuda版本和pytorch的cuda版本是一致的吗?因为有可能你pytorch用的cuda是conda装上去的libcudatoolkit,而你用来编译的是系统里的cuda. 你可以通过以下方式确定他们的版本: conda: ``` conda list | grep cuda ``` 系统cuda: ``` ls -alF /usr/local | grep cuda ```
@luweiqing 其实只需要保证所有的cuda都是一个版本即可。 对于3090的问题,只需要将所有的cuda保持在cuda11以上就可以了(这个repo中的模型就是在3090上训出来的)。 如果你系统中的是cuda 11.3,可以确定用到的一定是这个版本吗?比如你可以使用 ``` which nvcc ``` 来查看一下路径是不是在11.3的cuda中
@hitbuyi There is a `pretrained` flag during network initialization. You can turn it down here: https://github.com/cfzd/Ultra-Fast-Lane-Detection-v2/blob/f666202d39aedcc624248e65dd9c0604b1c6ac8c/model/model_tusimple.py#L2-L3
@pwshen Good point! It is correct. In short, the transform of resize and crop is learned inherently by the network. In fact, what we always want is the lane coordinates...
@wjywjy14 因为我不知道你是怎么把curvelane转换为tusimple的,所以我没法知道这个错误的根源。但是似乎错误提示是`len(h_samples[i][valid]`或者`len(lanes[i][j][valid])`里的变量是个int,所以你不能使用`len`函数
@walkmanyang 你可以尝试降低学习率
@Vergissmeinnic 首先,在out of domain上测试的结果是没有保证的。 其次,如果你想测试单张图,请参考`demo.py`.
@Durobert The values of `1640./25` and `32./534*590` are the offset of test-time-augmentation (TTA). During TTA, we would first shift the image, and get the prediction of the shifted image. Then...
@Durobert It should be correct. In fact, another interesting point is that: if you always do TTA both in the opposite directions with the same shift, you can directly average...
@Durobert The offset with the crop operation is a little tricky, and sorry I have forgotten the derivation details. However, the core idea is the same, and it is just...