Abstract In this work we present InsideNet , a novel tool built on top of Caffe DL framework aimed to assist researchers in exploring the values generated during the inference procedure of a Convolutional Neural Network (CNN). More precisely, InsideNet allows in-depth analysis of the values of the filters and fmaps within the convolution layers of a trained CNN during an ongoing inference procedure. To do so, InsideNet features three main operation modes. First, the Fmap Visualization Mode (FVM), which allows users to examine in a visual manner the generated fmap channels during the inference procedure of a set of images. Second, the Statistic Collector Mode (SCM), which offers a rich set of statistics for the fmap channels and weights of every convolutional layer. And third, the Histogram Collector Mode (HCM), which allows deeper exploration of value-based patterns by generating the corresponding histograms of the network. In addition, we describe a methodology and a set of standard metrics that can be used to characterize any CNN using InsideNet . We demonstrate the potential of InsideNet by applying it to two real case studies that focus on contemporary lightweight CNNs. The first is a comprehensive characterization of the novel Google’s MobileNets CNN, aimed to find out new patterns that can lead to new hardware optimizations. Specifically, we have found that on average, each fmap could be divided into 8 parts of the same size, each of which with consecutive, identical values. The second case study uses InsideNet for understanding and solving the accuracy drop (around 6%) that appears in SqueezeNet CNN when it is quantized from 32 to 16 bits.