🌦️ Weather Classification using CNN & Transfer Learning

This project classifies images of different weather conditions into 11 categories:
dew, fogsmog, frost, glaze, hail, lightning, rain, rainbow, rime, sandstorm, snow.

---

📂 Dataset

• Custom dataset stored in ./data/weather class/dataset/
• Each folder corresponds to one weather category
• No predefined train/test split → split done using validation_split in TensorFlow

---

🧠 Models Used

1. Custom CNN (Convolutional Neural Network)

   • 3 Conv2D layers + Dense layers
   • Accuracy: ~17% (not great)

2. MobileNetV2 (Transfer Learning)

   • Pretrained on ImageNet
   • Fine-tuned for weather dataset
   • Accuracy: ~80–85% after tuning

---

📊 Results

Model	Accuracy
Custom CNN	~17%
MobileNetV2	~80–85%

---

🚀 How to Run

1. Clone the repository

   git clone https://github.com/Vision-Ali/Weather-Classification-CNN.git
   cd Weather-Classification-CNN