Pruning-and-Quantization

This project involves the implementation of the Pruning and Quantization of VGG16 Network for Image Classification of CIFAR10. The pruning techniques used are fine-grained pruning and channel pruning. K-means quantization is implemented and applied to the model after that. To see the detailed working of these model optimization techniques, check the code here.

Network Architecture

Pruning Visualization

Fine Grained Pruning

Before Pruning	After Pruning

Quantization Visualization

K-means Quantization