Ncnn: pytorch转onnx之后再转ncnn: Split not supported yet! # axis=1 # split=7
我的模型结构中使用到了torch.split对channel进行切割，转换到onnx没有问题，但是通过onnx转换到ncnn出错，请问split操作是支持的嘛？或者有什么办法可以替换这个操作？
hackeritchy
All 7 comments
ncnn 中split操作是整个blob复制，没有做channel切割，真正做channel切割的是slice层，但不支持onnx::split 到 ncnn::slice的转换，可以pytorch > caffe > ncnn ，或者改写onnx2ncnn.cpp 添加split to slice操作，我改了一版但是slice层好像有bug ，一到这层就崩
hanfanggithub
on 16 Apr 2019
了解，我看看w
nihui
on 16 Apr 2019
@nihui @hanfanggithub 请问这个问题有解决方案了嘛
hackeritchy
on 24 Apr 2019
@hanfanggithub I am stuck at the same problem.
@nihui overriding split to slice is buggy. Can we expect a proper split operation in onnxtoncnn
or any work around?
soham24
on 21 Jun 2019
@hanfanggithub did you found solution to this?
soham24
on 22 Jun 2019
@soham24 You may be able to try the following https://github.com/biubug6/ncnn_faceboxes/tree/master/tools/onnx
biubug6
on 28 Jul 2019
https://github.com/Tencent/ncnn/commit/1b910efea52621e36813874334d18920c48ec6d4
slice op should be converted to ncnn crop properly now
nihui
on 2 Aug 2019
Related issues
AlexeyAB
·
3Comments
SisterL
·
3Comments
baiyecha
·
3Comments
yizhaoyanbo
·
3Comments
xhappy
·
5Comments
Most helpful comment
@soham24 You may be able to try the following https://github.com/biubug6/ncnn_faceboxes/tree/master/tools/onnx