Caffe-Based Dockerized Implementation and Network Anatomy Based Benchmark Analysis of Popular ImageNet Classification Deep CNN Architectures. Deep Convolutional Neural Networks (CNNs) have recently demonstrated the state-of-the-art classification performance on ImageNet Large Scale Visual Recognition Challenge (ILSVRC) since 2012, yet there is relatively no clear understanding of why they perform so well, or how they might be improved, In this paper we present a novel benchmarking studies of popular CNN architectures that have been listed as winners in the ILSVRC challenge between 2012 and 2016, The benchmarking was carried out from epochs timing, memory consumption, and network parameters visualization performed on a group of CNN architectures using Caffe Deep Learning framework running on a Docker container with random test samples, then the results were investigated and interpreted based on the networks anatomy and their hyper-parameters selection