https://github.com/puureya2/machine-learning-tensorflow

Year 2, Principles of Artificial Intelligence ISB42403, Final Project, TensorFlow-Keras CNN Model Training, Machine Learning
https://github.com/puureya2/machine-learning-tensorflow

cnn-model keras matplotlib python tensorflow

Last synced: about 2 months ago
JSON representation

Year 2, Principles of Artificial Intelligence ISB42403, Final Project, TensorFlow-Keras CNN Model Training, Machine Learning

• Host: GitHub
• URL: https://github.com/puureya2/machine-learning-tensorflow
• Owner: puureya2
• Created: 2025-01-21T06:40:49.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2025-07-21T08:01:56.000Z (12 months ago)
• Last Synced: 2025-07-21T10:11:29.855Z (12 months ago)
• Topics: cnn-model, keras, matplotlib, python, tensorflow
• Language: Python
• Homepage:
• Size: 257 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# 🌾 Crops Sorter Model

A machine learning pipeline for detecting crop health patterns using **Convolutional Neural Networks (CNNs)**. The system is trained on 10,000+ labeled agricultural images to predict crop conditions with high accuracy, and includes analytics and visualizations of model performance.

---

## 🧠 Core Features

- 🤖 Trained 3 custom CNN architectures using **TensorFlow** and **Keras**

- 📊 Achieved **90.07% accuracy** on validation data

- 📈 Designed analytics scripts to evaluate 4 key model metrics:

  - Accuracy

  - Precision

  - Recall

  - Loss

- 🌐 Visualized predictions using **Matplotlib** with bar graphs and cluster charts

---

## 🛠️ Tech Stack

| Purpose                 | Technology         |

|-------------------------|--------------------|

| Model Training          | TensorFlow, Keras  |

| Data Preprocessing      | Scikit-learn       |

| Data Visualization      | Matplotlib         |

| Language                | Python             |

---

## 🧪 Training Overview

- Dataset: >10,000 images across various crop-health categories  

- Preprocessing: Normalization, augmentation, and one-hot encoding  

- CNN Variants:

  - ResNet-inspired shallow net

  - Custom-built 6-layer CNN

  - Lightweight MobileNetv2 baseline  

- Evaluation: Accuracy calculated via validation split (90.07%)

---

## 📊 Visualizations

Using `matplotlib`, the following charts were generated:

- **Bar graphs** comparing model performance

- **Cluster plots** to group prediction categories

- Accuracy/Loss trends over training epochs

---

## 📁 Files

- `train_model.py` — Model architecture and training loop

- `evaluate.py` — Script to calculate and compare model metrics

- `visualize.py` — Generates performance graphs

- `README.md` — Project documentation

---

## 🧠 Future Improvements

- Integrate early stopping and learning rate schedulers

- Expand dataset with underrepresented crops

- Deploy trained model to a web/mobile interface for farmer use

---

## 📅 Timeline

**January – February 2025**  

Created as a solo AI-agriculture capstone to explore the intersection of deep learning and food security.

---

## 📬 Contact

**Kevin Chifamba**  

📧 kevinnanashe@gmail.com  

🔗 [LinkedIn](https://www.linkedin.com/in/yourprofile) • [GitHub](https://github.com/your-username)

---