tensorflow_inception_graph

tensorflow_inception_graph/README.md

@@ -27,19 +27,9 @@ net = cv2.dnn.readNet("tensorflow_inception_graph_2026jul.onnx")
 ```
 
 ### C++
-The C++ demo runs inference with OpenCV's DNN module. Adjust the OpenCV paths to your setup:
 ```bash
-OCV=/path/to/opencv                          # OpenCV source tree
-OCVBUILD=/path/to/opencv/build               # OpenCV build directory (generated headers + libs)
-g++ -std=c++17 demo.cpp -o demo \
-  -I$OCV/include \
-  -I$OCV/modules/core/include \
-  -I$OCV/modules/dnn/include \
-  -I$OCV/modules/imgproc/include \
-  -I$OCV/modules/imgcodecs/include \
-  -I$OCVBUILD \
-  -L$OCVBUILD/lib -Wl,-rpath,$OCVBUILD/lib -lopencv_dnn -lopencv_imgcodecs -lopencv_imgproc -lopencv_core
-./demo --model tensorflow_inception_graph_2026jul.onnx --image example_outputs/input_image.png
+cmake -B build && cmake --build build
+./build/demo --model tensorflow_inception_graph_2026jul.onnx --image example_outputs/input_image.png
 ```
 
 ## Conversion

tensorflow_inception_graph/demo.cpp

@@ -1,13 +1,13 @@
 #include <opencv2/dnn.hpp>
 #include <opencv2/imgproc.hpp>
 #include <opencv2/imgcodecs.hpp>
-#include <algorithm>
-#include <array>
 #include <fstream>
 #include <iostream>
 #include <string>
 #include <vector>
 
+using namespace cv;
+
 static std::string argVal(int argc, char** argv, const std::string& key, const std::string& def)
 {
     for (int i = 1; i + 1 < argc; ++i)
@@ -22,44 +22,45 @@ int main(int argc, char** argv)
     std::string output = argVal(argc, argv, "--output", "example_outputs/output_image.png");
     std::string labels = argVal(argc, argv, "--labels", "");
 
-    cv::Mat img = cv::imread(image);
+    Mat img = imread(image);
     if (img.empty())
     {
         std::cerr << "could not read image: " << image << std::endl;
         return 1;
     }
 
-    cv::Mat rgb;
-    cv::cvtColor(img, rgb, cv::COLOR_BGR2RGB);
-    cv::resize(rgb, rgb, cv::Size(224, 224));
+    Mat rgb;
+    cvtColor(img, rgb, COLOR_BGR2RGB);
+    resize(rgb, rgb, Size(224, 224));
     rgb.convertTo(rgb, CV_32F);
-    if (!rgb.isContinuous()) rgb = rgb.clone();
 
-    int blobShape[] = {1, 224, 224, 3};
-    cv::Mat blob(4, blobShape, CV_32F, rgb.data);
-    cv::dnn::Net net = cv::dnn::readNetFromONNX(model);
+    int dims[] = {1, 224, 224, 3};
+    Mat blob(4, dims, CV_32F, rgb.data);
+
+    dnn::Net net = dnn::readNet(model);
     net.setInput(blob);
-    cv::Mat scoresMat = net.forward();
-    float* scores = (float*)scoresMat.data;
-    int n = (int)scoresMat.total();
-    int top = (int)(std::max_element(scores, scores + n) - scores);
-    float conf = scores[top];
+    Mat scores = net.forward().reshape(1, 1);
 
-    std::string label = std::to_string(top);
+    Point classId;
+    double conf;
+    minMaxLoc(scores, 0, &conf, 0, &classId);
+
+    std::string label = std::to_string(classId.x);
     if (!labels.empty())
     {
         std::ifstream f(labels);
         std::vector<std::string> names;
         std::string line;
         while (std::getline(f, line)) names.push_back(line);
-        if (top < (int)names.size()) label = names[top];
+        if (classId.x < (int)names.size()) label = names[classId.x];
     }
-    std::cout << "class " << top << " " << label << " confidence " << conf << std::endl;
 
-    cv::Mat out = img.clone();
-    cv::putText(out, cv::format("%s (%.2f)", label.c_str(), conf), cv::Point(10, 30),
-                cv::FONT_HERSHEY_SIMPLEX, 1.0, cv::Scalar(0, 255, 0), 2);
-    cv::imwrite(output, out);
+    std::cout << "class " << classId.x << " " << label << " confidence " << conf << std::endl;
+
+    Mat out = img.clone();
+    putText(out, format("%s (%.2f)", label.c_str(), conf), Point(10, 30),
+            FONT_HERSHEY_SIMPLEX, 1.0, Scalar(0, 255, 0), 2);
+    imwrite(output, out);
     std::cout << "wrote " << output << std::endl;
     return 0;
 }

tensorflow_inception_graph/demo.py

@@ -3,6 +3,7 @@ import os
 
 import cv2 as cv
 import numpy as np
+import onnxruntime as ort
 
 here = os.path.dirname(os.path.abspath(__file__))
 
@@ -21,9 +22,8 @@ def main():
 
     rgb = cv.resize(cv.cvtColor(img, cv.COLOR_BGR2RGB), (224, 224)).astype(np.float32)
 
-    net = cv.dnn.readNetFromONNX(args.model)
-    net.setInput(rgb[None])
-    scores = net.forward().ravel()
+    sess = ort.InferenceSession(args.model, providers=["CPUExecutionProvider"])
+    scores = sess.run(None, {sess.get_inputs()[0].name: rgb[None]})[0].ravel()
 
     top = int(np.argmax(scores))
     conf = float(scores[top])