fix pre-processing

Signed-off-by: jagadeesh <[email protected]>

cpp/src/examples/image_classifier/resnet-18/resnet-18_handler.cc

@@ -71,45 +71,44 @@ std::vector<torch::jit::IValue> ResnetHandler::Preprocess(
 
         cv::Mat image = cv::imdecode(data_it->second, cv::IMREAD_COLOR);
 
-        cv::cvtColor(image, image, cv::COLOR_BGR2RGB);
-
         // Check if the image was successfully decoded
         if (image.empty()) {
           std::cerr << "Failed to decode the image." << std::endl;
         }
 
-        // Resize
-        const int newWidth = 256, newHeight = 256;
-        cv::Mat resizedImage;
-        cv::resize(image, resizedImage, cv::Size(newWidth, newHeight));
-
         // Crop image
-        const int cropSize = 224;
-        const int offsetW = (resizedImage.cols - cropSize) / 2;
-        const int offsetH = (resizedImage.rows - cropSize) / 2;
+        const int rows = image.rows;
+        const int cols = image.cols;
+
+        const int cropSize = std::min(rows, cols);
+        const int offsetW = (cols - cropSize) / 2;
+        const int offsetH = (rows - cropSize) / 2;
 
         const cv::Rect roi(offsetW, offsetH, cropSize, cropSize);
-        cv::Mat croppedImage = resizedImage(roi).clone();
+        image = image(roi);
 
-        // Convert the OpenCV image to a torch tensor
-        // Drift in cropped image
-        // Vision Crop: 114, 118, 115, 102, 106, 97
-        // OpenCV Crop: 113, 118, 114, 100, 106, 97
-        torch::TensorOptions options(torch::kByte);
-        torch::Tensor tensorImage = torch::from_blob(
-            croppedImage.data,
-            {croppedImage.rows, croppedImage.cols, croppedImage.channels()},
-            options);
+        // Resize
+        cv::resize(image, image, cv::Size(224, 224));
 
+        // Convert BGR to RGB format
+        cv::cvtColor(image, image, cv::COLOR_BGR2RGB);
+
+        image.convertTo(image, CV_32FC3, 1 / 255.0);
+
+        // Convert the OpenCV image to a torch tensor
+        torch::Tensor tensorImage = torch::from_blob(image.data, {image.rows, image.cols, 3}, c10::kFloat);
         tensorImage = tensorImage.permute({2, 0, 1});
-        tensorImage = tensorImage.to(torch::kFloat32) / 255.0;
+        tensorImage.unsqueeze_(0);
 
         // Normalize
-        torch::Tensor normalizedTensorImage =
-            torch::data::transforms::Normalize<>(
-                {0.485, 0.456, 0.406}, {0.229, 0.224, 0.225})(tensorImage);
-        normalizedTensorImage.clone();
-        batch_tensors.emplace_back(normalizedTensorImage.to(*device));
+        std::vector<double> norm_mean = {0.485, 0.456, 0.406};
+        std::vector<double> norm_std = {0.229, 0.224, 0.225};
+
+        tensorImage =
+          torch::data::transforms::Normalize<>(norm_mean, norm_std)(tensorImage);
+
+        tensorImage.clone();
+        batch_tensors.emplace_back(tensorImage.to(*device));
         idx_to_req_id.second[idx++] = request.request_id;
       } else if (dtype_it->second == "List") {
         // case3: the image is a list