Fruit Recognition Quantity and Quality Estimation Using Robotic Vision System

This innovative project aimed at revolutionizing the fruit industry. By leveraging robotic vision systems, I have developed a solution that automates fruit recognition, quantity estimation, quality assessment, and disease detection processes. Utilizing advanced technologies such as Python, TensorFlow, and the Fruits 360 open-source dataset, I have achieved remarkable accuracy in my operations.

Prototype Product

Objectives

• Fruit Recognition: Identifying the type and quantity of fruits.
• Quality Assessment: Assessing the quality of fruits.

Features

• Image Acquisition: I utilize a Pi camera to capture images of the fruits.
• Image Pre-processing: Enhancing images through color conversion and filtering.
• Feature Extraction: Extracting texture and color features to characterize fruits.
• Classification: Employing machine learning algorithms for accurate classification.

Block Diagram

!

Block Diagram Explanation

• Stage I (Input Image): I capture input images using the Pi camera.
• Stage II (Preprocessing): I perform color conversion and filtering to enhance image quality.
• Stage III: I conduct color enhancement and morphological operations.
• Stage IV: I extract features using AlexNet and perform classification.
• Output: Displaying cultivation status, fruit quality, and quantity on a monitor.

Software and Hardware Requirements

Software Requirements

• Raspbian
• Python
• Keras
• TensorFlow

Hardware Requirements

• Raspberry Pi
• Pi-Camera
• Power cable
• Memory card
• Connecting cables
• Pi-protection case
• Temperature Sensor
• Humidity Sensor

Advantages

• Efficiency: Speeding up the fruit assessment process, reducing time.
• Accuracy: Improving accuracy and efficiency.
• Adaptability: Suitable for varying illuminant conditions.

Output Samples

Sample Output of Apple:

Sample Output of Multiple Apples:

Sample Output of Diseased Banana:

Conclusion

This Project offers a comprehensive solution for fruit recognition, assessment, and disease detection. By closely monitoring environmental conditions and fruit quality, I ensure optimal fruit production and quality control.

Applications

• Food industry for segregating healthy and diseased fruits.
• Agriculture sector for harvesting good and healthy fruits.
• Marketing and export industries for quality assessment and grading.

Future Enhancements

• Fine-tuning trained data for increased prediction accuracy.
• Incorporating additional features to enhance classification.
• Developing a user-friendly mobile application interface for enhanced accessibility.