Performance Analysis of Model Optimization Techniques using Pre-Trained CNN Models

An exploration into the domain of image classification for automobile parts, employing Convolutional Neural Networks (CNNs) and the Inception V3 architecture. The study systematically investigates the effectiveness of different optimization techniques, including Adam and Stochastic Gradient Descent (SGD), coupled with varied weight initialization methods and regularization techniques. The evaluation aims to discern optimal configurations that enhance model performance. The research sheds light on the relevance and efficiency of specific optimization strategies for CNNs in image classification tasks related to automobile parts, offering insights for practitioners and researchers in the field of deep learning.