TetrisAI: a new way to play Tetris
summary
A playable Tetris variant maps wrist positions from a lightweight pose-estimation model into left/right/rotate/drop zones using OpenCV and Pygame, records MP4s to Amazon S3, and posts line-clear scores (100/300/700/1500 for 1–4 lines) to DynamoDB shown on a Streamlit leaderboard. Two-player mode communicates via MQTT over AWS IoT Core with a single-use “bomb” message that randomly clears blocks; models are converted to TensorRT and run in real time on an NVIDIA Jetson Xavier NX (games ~180s), while OpenCV handles camera streams and video capture.
AI-written summary · about
chapters (5)
related talks
37mPython, Hardware, Networks and your toolbox for Interactive Art.
20mAI for Gaming: How I Built a Bot to Play a Video-Game with R and Python
05mIncredible things your community is creating with TensorFlow.js
20mTetris: Memory-efficient Serverless Inference through Tensor Sharing
46mAI Takeover
50mBuilding a Pong playing AI in just 1 hour (plus 4 days training...)
39mArtificial Intelligence vs. Natural Stupidity
22mStrike a Pose: Gesture Recognition in JS with Machine Learning