Deep-learning-from-scratch-

• Feed forward Neural Networks, Convolutional Neural Networks, Recurrent Neural Networks written using numpy library including Pooling layers, Activation functions, Optimization layers, Batch normalization from scratch.

Solar Panel Image Project

• Multi-class classification problem
• Dataset with solar panels images containing cracks, inactive regions, cracks and inactive regions
• Pretraining with ResNet architecture
• Hyperparameter tuning, Global average pooling, Drop out layers
• Checkpoints
• Goal : Reach F1 score > 0.6
• Reached a F1 Score of 0.87

[2023]