torch2trt
torch2trt copied to clipboard
MaskRCNN
Hello, is there support for MaskRCNN? I am using the torchvision.models.detection.maskrcnn_resnet50_fpn(pretrained=True) model, but get an error when trying to convert to torch2trt. Thanks!
I also get this:
Warning: Encountered known unsupported method torch.Tensor.__iter__ Warning: Encountered known unsupported method torch.Tensor.__iter__ Warning: Encountered known unsupported method torch.as_tensor Warning: Encountered known unsupported method torch.as_tensor
I also get an error of [TensorRT] ERROR: (Unnamed Layer* 1033) [ElementWise]: elementwise inputs must have same dimensions or follow broadcast rules (input dimensions were [1,256,2,2] and [1,2,2,76]). Are the plans to get torch2trt to work for this; and if not, what would be the steps for it? Thanks!
Hi, Did you solve the issue?
Hi, no. I am using https://github.com/onnx/onnx-tensorrt
That's what I thought, but did you have to customize new layers or it was a plain and simple conversion Pytorch->Onnx->TensorRT?
I didn't have to do that much work for conversion.
@felipevw @peasant98 Could you give link of any guidance for Pytorch->Onnx->TensorRT conversion?
@DuyguSerbes Did you figure it out?
any update?