Chapter 16: Capstone Projects

This chapter outlines potential capstone project ideas that leverage the concepts and techniques discussed throughout this documentation. These projects offer practical applications and opportunities to deepen your understanding.

Project Ideas

Here are several capstone project ideas, categorized by their primary focus area:

1. Defect Detection in Manufacturing

Description: Develop a system to automatically identify defects in manufactured products using computer vision. This could involve analyzing images of parts, components, or finished goods to detect anomalies, imperfections, or deviations from expected quality standards.

Potential Scope:

• Data Acquisition: Capture images of products using cameras (e.g., industrial cameras, webcams).
• Preprocessing: Implement image enhancement techniques like noise reduction, contrast adjustment, and resizing.
• Feature Extraction: Utilize methods like edge detection, texture analysis, or deep learning features.
• Defect Classification/Localization: Employ machine learning models (e.g., Support Vector Machines, Convolutional Neural Networks) to classify whether a defect is present and, if so, where it is located.
• Reporting: Generate reports or alerts when defects are detected.

Example Application: Detecting scratches on metal surfaces, identifying missing components on circuit boards, or finding cracks in molded plastic parts.

2. Document Workflow Automation

Description: Design and implement a system to automate repetitive tasks associated with document processing. This could involve extracting information from documents, categorizing them, routing them to appropriate workflows, or generating new documents based on predefined templates.

Potential Scope:

• Document Ingestion: Support various document formats (PDF, images, text files).
• Optical Character Recognition (OCR): Extract text from scanned documents.
• Information Extraction: Identify and extract key data fields (e.g., names, dates, invoice numbers) using techniques like regular expressions, named entity recognition, or template matching.
• Document Classification: Automatically assign documents to predefined categories.
• Workflow Integration: Connect with other systems or trigger actions based on document content and type.

Example Application: Automating invoice processing, digitizing and categorizing customer feedback forms, or streamlining the onboarding process by extracting information from employee documents.

3. License Plate Recognition (LPR)

Description: Build a system that can automatically detect and read license plates from images or video streams. This technology is widely used in traffic management, parking enforcement, and security applications.

Potential Scope:

• Vehicle Detection: Identify vehicles within an image or video frame.
• License Plate Detection: Locate the license plate region on the vehicle.
• Character Segmentation: Isolate individual characters on the license plate.
• Character Recognition: Identify each character using OCR or specialized models.
• Result Formatting: Combine recognized characters into a readable license plate string.

Example Application: Automating toll collection, managing access to parking lots, or identifying vehicles of interest in surveillance footage.

4. Retail Analytics

Description: Develop solutions to gain insights into customer behavior and store operations by analyzing data collected in retail environments. This category includes specific applications like people counting and shelf monitoring.

4.1. People Counting

Description: Implement a system to accurately count the number of people entering or present in a specific area, such as a store entrance or a particular aisle.

Potential Scope:

• Camera Feed Processing: Analyze video streams from cameras.
• Person Detection/Tracking: Identify and track individuals within the frames.
• Counting Logic: Implement algorithms to increment/decrement counts based on entry/exit points or presence within a defined zone.
• Real-time Display/Reporting: Visualize current counts and historical trends.

Example Application: Measuring store foot traffic, optimizing staffing based on occupancy, or understanding customer flow patterns within a store.

4.2. Shelf Monitoring

Description: Create a system to monitor product availability and placement on store shelves. This can help in identifying out-of-stock items, misplaced products, and planogram compliance.

Potential Scope:

• Shelf Image Analysis: Capture images of store shelves.
• Product Recognition: Identify specific products on the shelves.
• Stock Level Estimation: Determine the quantity of each product available.
• Anomaly Detection: Flag empty shelves, misplaced items, or products not in their designated locations.

Example Application: Alerting staff to replenish low-stock items, identifying products that are not in their correct shelf positions, or ensuring promotional displays are correctly set up.