Client Background

Our client, one of the largest pipe manufacturers globally, operates a big facility in Texas, USA. With hundreds of racks in an open yard warehouse, they manually counted pipes for auditing and tracking purposes using drone footage. The manual counting process was time-consuming, labor-intensive, and prone to errors. The client sought an automated solution to enhance accuracy and efficiency.

Objective

The primary goal was to develop an automated system to count pipes accurately using drone footage and display the results on a dashboard. Additionally, the system needed to provide a downloadable spreadsheet of the counting data and integrate with the client's ERP system in future phases.

Solution Design

1. Data Collection and Model Training:

• Drone Footage Collection: We gathered necessary drone videos capturing various conditions, angles, and distances to ensure a robust dataset.
• Image Extraction: Extracted frames from the videos to create a comprehensive image dataset.
• AI Model Training: Trained a computer vision AI model using these images to accurately count pipes. The model was designed to adapt to the dynamic conditions of the drone footage.

2. Dashboard Development:

• Real-Time Counting: Developed a real-time counting dashboard where the AI algorithm processes the video footage to count pipes.
• Region Selection Tool: Implemented a flexible tool allowing users to draw a region on the running video. The AI algorithm focuses on this region to provide accurate counting.
• Historical Data: Integrated a feature to search and display historical counting data.

3. Downloadable Reports:

• Designed the system to generate downloadable spreadsheets of the counting data for audit and reporting purposes.

Challenges and Solutions

Dynamic Drone Footage:

• Challenge: The drone camera’s constant movement, varying distances, and changing angles made it difficult to maintain consistent counting accuracy.
• Solution: The AI model was trained on diverse conditions to handle these variations. The region selection tool allowed users to specify areas of interest, improving accuracy despite dynamic footage.

Results

• Enhanced Accuracy: The AI-powered counting system provided significantly more accurate counts compared to manual methods.
• Increased Efficiency: Reduced the time and labor required for pipe counting, allowing employees to focus on other critical tasks.
• User-Friendly Interface: The flexible dashboard and region selection tool empowered users to interact with the system intuitively.
• Data Accessibility: Historical data search and downloadable reports enhanced transparency and ease of access to counting data.

Next Steps

The successful POC paved the way for integrating the counting system with the client's ERP, enabling end-to-end automation of the pipe tracking and auditing process. Future enhancements will focus on improving AI model accuracy under even more varied conditions and expanding the system’s capabilities to other parts of the facility.

Conclusion

This case study demonstrates the potential of AI-driven solutions in transforming manual, error-prone processes into efficient, automated systems. Our flexible and robust pipe counting system not only met the client’s immediate needs but also set the stage for further technological integration and process automation.