I have to do backbone inference in TensorRT, but the post process is using ... out = torch::from_blob(outputs[i], {1, num, dim, dim}, ...
確定! 回上一頁