https://github.com/thepredictivedev/dino-playing-rl-agent-using-dqn

Using screen captured frames from the computer as input, we train an RL agent using the DQN algorithm with the CNN policy to play the Dino Game
https://github.com/thepredictivedev/dino-playing-rl-agent-using-dqn

Last synced: 4 months ago
JSON representation

Using screen captured frames from the computer as input, we train an RL agent using the DQN algorithm with the CNN policy to play the Dino Game

• Host: GitHub
• URL: https://github.com/thepredictivedev/dino-playing-rl-agent-using-dqn
• Owner: ThePredictiveDev
• Created: 2024-01-27T14:29:37.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-01-27T14:39:41.000Z (over 1 year ago)
• Last Synced: 2025-01-27T23:44:54.780Z (5 months ago)
• Language: Jupyter Notebook
• Size: 25.4 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Dino-AI: Dino Playing RL Agent

An agent trained using the principles of SPTDL using DQN RL algorithm on a CNN.

## Environment:

The environment is a custom gym environment made using the base gym environment class

Information about the observation space is obtained using MSS library as frames are given as input to the CNN

## Preprocessing:

Preprocessing is done using OpenCV library

## Algorithm (Model)

Model is obtained from Stable-Baselines3 library

### All but one libraries used are upto date (OpenAI Gym is used not Gymnasium)