aircraft-classification-densenet201

🛩️ Aircraft Classification Model

This model is based on DenseNet201 architecture, fine-tuned on the FGVC Aircraft (10,000 images) dataset.
It can classify aircraft into different variants with high accuracy.

---

📚 Dataset

• Name: FGVC Aircraft Dataset
• Images: 10,000+
• Source: FGVC Aircraft

---

🏗️ Model Architecture

• Base Model: DenseNet201 (pre-trained on ImageNet)

• Custom Layers:

  • GlobalAveragePooling2D
  • Dense(1024) + Mish activation + BatchNorm + Dropout
  • Dense(512) + Mish activation + BatchNorm + Dropout
  • Final Dense layer with Softmax activation

• Optimizer: Adam

• Learning Rate Schedule: CosineDecay

• Batch Size: 8

• Epochs: 50

---

📈 Performance

• Achieved high validation accuracy.
• Used EarlyStopping, ReduceLROnPlateau, and ModelCheckpoint for optimal training.

---

🛠️ How to Use

import tensorflow as tf

# Load the model
model = tf.keras.models.load_model('aircraft_model_finetuned.h5', custom_objects={"mish": mish})

# Predict
prediction = model.predict(preprocessed_image)