Implementation of the CartPole from OpenAI's Gym using only visual input for Reinforcement Learning control with DQN