Single image based pathfinding in virtual environments this is a part of my bachelorthesis Steps Collect data from the virtual environment ViZDoom Build and train a CNN Images