Project

Problem:

• Manual apple grading is time consuming and inconsistent

Goal:

• Create a system that detects apples, estimates size, color, and grades in real time
• Produce a low-cost solution for small scale apple growers
• Used at USU Kaysville farm in apple orchards

System

Methods

• Camera to take pictures of apples on conveyer belt
• Control background and lighting to reduce shadows & glare
• Images processed using color levels
• Filters used to mask background
• Apples detected with size & color estimates
• Apples sorted into bins based on predefined values

Conclusion

Results:

• Successfully estimated apple size and color for apple grading
• Showed that automated apple grading can be accurate and cost effective on a small scale
• System works best with consistent lighting and background

Future Improvements:

• Handling overlapping apples in real-time
• Expanding system to detect defects such as bruises and blemishes