Figure 1. The layout of our GitHub repository.

A configuration file is generated in the first step, which maintains directory paths and parameter values used by subsequent modules. Tracking happens in two steps—first, we need to train the network on training dataset; second, object detection is done using the trained CNN on the image. Each step here is a separate module that can be run by users. Black arrows represent the directional flow of executable files. Blue arrows represent input/output flow of data in the modules.