Skip to content

README.md

Latest commit

 

History

History
123 lines (99 loc) · 3.52 KB

README.md

File metadata and controls

123 lines (99 loc) · 3.52 KB

Automated-Harvest-Machine-for-Precision-Framing

This project aims to develop an Automated Harvest Machine for precision farming by integrating software and hardware solutions. The software employs image processing and machine learning to identify the ripeness of the crops. Based on this analysis, the hardware system accurately harvests the crop improving efficiency, reducing manual labor.

  1. System Overview Purpose: Automatically classify areca nuts into: 🟠 Ripe (Orange/Yellow) 🟢 Unripe (Green) ⚫ Rotten (Black/Brown) ❌ Not Valid (Other objects)

  2. Technical Components

A. Core Files File Purpose main.py Main menu & user interface capture_image.py Webcam image capture preprocess.py Image feature extraction train_model.py Machine learning model training predict.py Classification logic webcam_classify.py Real-time classification

B. Key Technologies OpenCV: Image processing Scikit-learn: Random Forest classifier NumPy: Numerical operations Joblib: Model saving/loading

  1. Workflow

  2. Image Input Webcam capture OR test image Example: test_samples/unripe.jpg

2.Preprocessing python def preprocess_image(): # Converts to HSV color space # Calculates color histograms # Extracts dominant color # Returns 173-dimensional feature vector Key Step: Checks if ≥10% of pixels match areca nut colors

3.Classification Uses trained Random Forest model Fallback to color rules if model unavailable python if 5° ≤ hue ≤ 25°: return "ripe" elif 35° ≤ hue ≤ 85°: return "unripe" else: return "rotten"

Output

=== Test Result === Prediction: unripe Confidence: 80.3% Details: Unripe (80.3%)

  1. Key Algorithms

A. Color Analysis HSV Color Space (Better than RGB for color detection) Hue (H): Color type (0-180° in OpenCV) Saturation (S): Color purity (0-255) Value (V): Brightness (0-255)

B. Machine Learning Random Forest Classifier 100 decision trees Trained on 173 features per image Outputs: [ripe_prob, unripe_prob, rotten_prob, not_valid_prob]

  1. Confidence Scores Confidence Range Interpretation ≥80% Very reliable 60-79% Moderately confident 40-59% Low confidence (marked "uncertain") <40% Rejected as "not_valid"

  2. Sample Output Explained Processing: test_samples/unripe.jpg Coverage: 96.24% # Excellent nut detection Dominant HSV: [175 188 135] # Greenish (H=175°), saturated (S=188) Prediction: unripe # Correct classification Confidence: 80.3% # High reliability

  3. Real-World Applications Quality Control in areca nut packaging Automated Sorting machines Farm Harvest Timing assistance

  4. Try It Yourself Run the system:

python main.py Test your own images by dropping them in test_samples/

🔧 Technical Details Feature Extraction Histogram bins: 50 (Hue), 60 (Saturation), 60 (Value) Dominant color: Median HSV values Coverage check: ≥10% areca-colored pixels required

Machine Learning Model Algorithm: Random Forest Classifier Parameters: 150 trees, class_weight="balanced" Features: 173 per image Accuracy: Typically 85-95% on validation set

📈 Performance Metrics Class Accuracy Typical Confidence Ripe 90-95% 85-98% Unripe 85-92% 80-95% Rotten 80-88% 75-92% Not_Valid 95-99% 90-100%

🤝 Contributing Fork the repository Create feature branch (git checkout -b feature/improvement) Commit changes (git commit -am 'Add new feature') Push to branch (git push origin feature/improvement) Create Pull Request