Click the blue text above to follow us
WeChat Official Account:OpenCV Academy
Follow us for more knowledge on computer vision and deep learning
Introduction to the Padim Model
Padim primarily generates feature vectors from a series of normal samples using a CNN network, calculates multiple variance Gaussian matrices of the feature vectors, and obtains the feature data distribution of normal samples. Then, for the input sample image, it calculates the Mahalanobis distance between its feature vector and the training-generated feature data distribution, thus achieving anomaly detection and localization.
PadimModel Training and ExportSupports ONNX format orOpenVINOformat, please see the article below.
[Engineering Practice] Anomalib Anomaly Detection from Training to Deployment
OpenVINO 2025 Inference Steps
OpenVINO 2025 is more concise and user-friendly compared to previous versions of the C++ SDK, and it supports dynamic modification of input dimension parameters.
Loading the Model and Creating Inference Requests
ov::CompiledModel compiled_model = ie.compile_model(onnxpath, "CPU");this->infer_request = compiled_model.create_infer_request();Getting All Output Layer Names
auto outputs = compiled_model.outputs();for (auto item : outputs) { std::cout << item.get_any_name() << std::endl; names.push_back(item.get_any_name());}Modifying Input Layer Dimension Information
ov::Tensor input_tensor = infer_request.get_input_tensor();auto input_shape = input_tensor.get_shape();input_shape[0] = 1;input_shape[1] = 3;input_shape[2] = input_h;input_shape[3] = input_w;input_tensor.set_shape(input_shape);Getting Output Layer Dimension Information and Data Types
std::cout << "element type: " << output2.get_element_type().to_string() << std::endl;std::cout << "shape info: " << output2.get_shape().to_string() << std::endl;Inference
cv::Mat blob = cv::dnn::blobFromImage(image, 1.0 / 255.0, cv::Size(input_w, input_h), cv::Scalar(), true, false);memcpy(input_tensor.data<float>(), blob.ptr<float>(), data_s * sizeof(float));// Inference and return resultsthis->infer_request.infer();Getting Inference Output Data
this->infer_request.infer();auto output0 = this->infer_request.get_tensor(names[0]);const float* score = (float*)output0.data();std::cout << "anomaly score: " << score[0] << std::endl;auto output1 = this->infer_request.get_tensor(names[1]);const int* pred_label = (int*)output1.data();std::cout << "predict label: " << pred_label[0] << std::endl;Code Demonstration
For anomaly detection models such as PatchCore, Padim, and EfficientAD, I have implemented a class wrapper that allows for the deployment of anomaly detection models with just a few lines of code. The calling code is as follows:
std::shared_ptr<PatchCoreDetector> detector(new PatchCoreDetector());detector->initConfig("D:/python/yolov5-7.0/padim/model.xml", 224, 224);cv::Mat image = cv::imread("D:/python/yolov5-7.0/breaksmall.png");detector->detect(image);cv::imshow("Input Image", image);cv::waitKey(0);cv::destroyAllWindows();The results are as follows:
Learn OpenVINO 2025Large Model Deployment DevelopmentScan the code to view directly
Essential skills for upper computer and machine vision developers
Systematically master OpenCV C# SDK and workflow engine development. The course includes dozens of common machine vision cases implemented with OpenCV, with over 3000+ lines of code, and all materials, including courseware and code, are available for download. Q&A support is provided! Purchase now to get group buying discounts:
Recommended Reading
OpenCV 4.8 + YOLOv8 Object Detection C++ Inference Demonstration
ZXING + OpenCV Build Open Source Barcode Detection Application
Strategy | A Good Method to Learn Deep Learning in Just Three Months
Implement TensorRT 8.6 C++ Deep Learning Model Deployment in Three Lines of Code
Practice | YOLOv8 + OpenCV Implement DM Code Positioning Detection and Analysis
Object Detection Bounding Box Loss – From IOU to ProbIOU
Must-See for Beginners | Five Misconceptions in Learning Deep Learning
