Machine Vision Deep Learning Software for Surface Defect Detection and OK/NG Classification for Multiple types of Metal Part (VC++ / OpenCV / MFC / Open Source)
Budget: ₹12,500 – ₹37,500 INR
Machine Vision Deep Learning Software for Surface Defect Detection and OK/NG Classification for Multiple types of Metal Part (VC++ / OpenCV / MFC / Open Source)________________________________________
Overview
We need software that can perform surface defect detection and OK/NG classification on metal parts using deep learning techniques. The application must be developed in Visual Studio 2022 (VC++), using OpenCV 4.7.0, with an MFC-based GUI.
This software should analyze images of metal components, identify any visible defects, classify the part as OK or Defective, and visually mark defect areas. It should also optionally support live image input from a USB3 or GigE Vision industrial camera.
________________________________________
Technology Stack
• IDE: Visual Studio 2022
• Language: VC++
• UI Framework: MFC (Microsoft Foundation Classes)
• Image Processing & Deep Learning: OpenCV 4.7.0 (DNN module or ONNX support)
________________________________________
Input Dataset
• 100/N labeled images of OK metal parts
• 100/N labeled images of Defective metal parts
• Dataset will be used to train and validate the defect detection model
________________________________________
Software Requirements
1. Defect Detection & Classification
• Classify each input image as OK or Defective
• Highlight and visualize detected defect areas (bounding boxes or contours)
• Use deep learning for defect identification (via OpenCV DNN / ONNX)
2. Graphical User Interface
• MFC-based GUI with:
o Image selection & loading
o Display of input and processed images
o Classification result (OK / NG)
o Visual overlay of defect areas
3. Batch Image Processing
• Load multiple images from a folder
• Classify and mark each image
• Export results in a CSV file (filename, result, defect coordinates if any)
• Optionally save marked images
4. Live Camera Input (Optional)
• Capture images from USB3/GigE Vision camera
• Preview, capture, and classify images in real time
________________________________________
Deliverables
• Fully functional Windows software (.exe)
• Visual Studio 2022 solution and complete source code
• Trained deep learning model(s) (ONNX or compatible)
• CSV output functionality for results
• Documentation:
o Setup & build instructions
o Usage guide
o Instructions for retraining the model with new images
________________________________________
Notes
• Entire software must be implemented in C++ (no Python)
• Must use open-source libraries (no commercial licensing)
• Application should be modular, maintainable, and scalable
• Performance should be optimized for industrial use
Overview
We need software that can perform surface defect detection and OK/NG classification on metal parts using deep learning techniques. The application must be developed in Visual Studio 2022 (VC++), using OpenCV 4.7.0, with an MFC-based GUI.
This software should analyze images of metal components, identify any visible defects, classify the part as OK or Defective, and visually mark defect areas. It should also optionally support live image input from a USB3 or GigE Vision industrial camera.
________________________________________
Technology Stack
• IDE: Visual Studio 2022
• Language: VC++
• UI Framework: MFC (Microsoft Foundation Classes)
• Image Processing & Deep Learning: OpenCV 4.7.0 (DNN module or ONNX support)
________________________________________
Input Dataset
• 100/N labeled images of OK metal parts
• 100/N labeled images of Defective metal parts
• Dataset will be used to train and validate the defect detection model
________________________________________
Software Requirements
1. Defect Detection & Classification
• Classify each input image as OK or Defective
• Highlight and visualize detected defect areas (bounding boxes or contours)
• Use deep learning for defect identification (via OpenCV DNN / ONNX)
2. Graphical User Interface
• MFC-based GUI with:
o Image selection & loading
o Display of input and processed images
o Classification result (OK / NG)
o Visual overlay of defect areas
3. Batch Image Processing
• Load multiple images from a folder
• Classify and mark each image
• Export results in a CSV file (filename, result, defect coordinates if any)
• Optionally save marked images
4. Live Camera Input (Optional)
• Capture images from USB3/GigE Vision camera
• Preview, capture, and classify images in real time
________________________________________
Deliverables
• Fully functional Windows software (.exe)
• Visual Studio 2022 solution and complete source code
• Trained deep learning model(s) (ONNX or compatible)
• CSV output functionality for results
• Documentation:
o Setup & build instructions
o Usage guide
o Instructions for retraining the model with new images
________________________________________
Notes
• Entire software must be implemented in C++ (no Python)
• Must use open-source libraries (no commercial licensing)
• Application should be modular, maintainable, and scalable
• Performance should be optimized for industrial use