Aircraft Crack Detection Model

I achieved 90.81% prediction accuracy on testing data by developing a CNN model utilizing Keras/TensorFlow for accurate classification of airplane surface integrity.

Key features:

• Developed convolutional neural network for classification of cracks, missing paint, and missing screw heads
• Processed and augmented training images to improve model robustness
• Implemented data preprocessing pipeline for handling diverse input images
• Optimized model architecture to balance accuracy and training speed