Different Outputs in Python and Torchscript C++

Hello,

I was experimenting with Torchscript and I seem to be getting different outputs in my Python code versus my C++ code. I am simply loading the ResNet18 model, and loading a sample image: https://raw.githubusercontent.com/pytorch/hub/master/dog.jpg

I exported the resnet18 model following these instructions: https://pytorch.org/tutorials/advanced/cpp_export.html

I am running on Windows 10 64-bit. For the C++ environment, I am running on Microsoft Visual Studio 2019. I am running the debug build of libtorch 1.4.0 for Visual Studio. I believe this is running fine, as I am able to load models and tensors to put CPU and GPU.

I believe I have an issue with my C++ code, but I am not sure where I have the error. Below is my code along with the outputs I get.

Python Code:

import torch
import torchvision.models as models
import urllib.request
from PIL import Image
from torchvision import transforms

def norm_chan(chan, mean, std):
    b = (chan - mean) / std
    return b

# Load resnet18 model
model = models.resnet18(pretrained = True)

# Download an example image from the pytorch website
url, filename = ("https://github.com/pytorch/hub/raw/master/dog.jpg", "dog.jpg")
urllib.request.urlretrieve(url, filename)

# sample execution (requires torchvision)
input_image = Image.open(filename)
preprocess = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(224),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])

input_tensor = preprocess(input_image)

import numpy as np
input_tensor_numpy = input_tensor.numpy()
input_batch = input_tensor.unsqueeze(0) # create a mini-batch as expected by the model


# move the input and model to GPU for speed if available
if torch.cuda.is_available():
    input_batch = input_batch.to('cuda')
    model.to('cuda')

with torch.no_grad():
    output = model(input_batch)

import numpy as np
print()
output_cpu = output[0].cpu()
output_numpy = output_cpu.numpy()
print(output_numpy[0:9])

C++ code:

#include <torch/script.h>
#include <torch/torch.h>
#include <ATen/Tensor.h>
#include <opencv2/opencv.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <iostream>
#include <memory>

int main(int argc, const char* argv[]) {


    // Load up GPU stuff
    torch::DeviceType device_type;
    if (torch::cuda::is_available()) {
        std::cout << "CUDA available! Training on GPU." << std::endl;
        device_type = torch::kCUDA;
    }
    else {
        std::cout << "Training on CPU." << std::endl;
        device_type = torch::kCPU;
    }
    torch::Device device(device_type);

    torch::jit::script::Module module;
    std::cout << "Attempting to load resnet model.." << std::endl;
    try {
        // Deserialize the ScriptModule from a file using torch::jit::load().
        module = torch::jit::load("C:\\PyTorchPictureTest\\Model\\traced_resnet_model.pt");
        std::cout << "Successfully loaded resnet model" << std::endl;

        module.to(at::kCUDA);
        std::cout << "Moved model to gpu" << std::endl;

        // load image and transform
        cv::Mat image;
        image = cv::imread("C:\\PyTorchPictureTest\\dog.jpg", 1);

        cv::cvtColor(image, image, 4); //had to change this part of the code because the cvtColor Enums did not seem to be working
        cv::Mat img_float;
        image.convertTo(img_float, CV_32F, 1.0 / 255);
        cv::resize(img_float, img_float, cv::Size(224, 224));

        auto img_tensor = torch::from_blob(img_float.data, { 1, 224, 224, 3 }).to(torch::kCUDA);
        img_tensor = img_tensor.permute({ 0, 3, 1, 2 });
        img_tensor[0][0] = img_tensor[0][0].sub(0.485).div(0.229);
        img_tensor[0][1] = img_tensor[0][1].sub(0.456).div(0.224);
        img_tensor[0][2] = img_tensor[0][2].sub(0.406).div(0.225);
        auto img_var = torch::autograd::make_variable(img_tensor, false);

        std::vector<torch::jit::IValue> inputs;
        inputs.push_back(img_var);
        torch::Tensor out_tensor = module.forward(inputs).toTensor();
        std::cout << out_tensor.slice(1, 0, 10) << '\n';
    }
    catch (const c10::Error & e) {
        std::cerr << "error loading the model\n";
        return -1;
    }

    std::cout << "ok\n";
}

Output in Python:

Output from C++:

I would appreciate any insight on this. Thanks in advance.

1 Like

Maybe you’re suffering the same problem as me.