LegoSorter

Identify LEGO parts from photos, detect their color, and sort them into storage boxes using AI.

Features

• Part recognition via Brickognize API
• Color detection using OpenCV k-means clustering
• AI sorting — classifies parts into boxes/cavities via an LLM (Cerebras / OpenAI-compatible)
• Mobile web UI — capture parts with your phone camera, review and save results
• CLI mode for testing with local images

Quick start

pip install -r requirements.txt
cp .env.example .env
# edit .env — at minimum set CEREBRAS_API_KEY

Web app

python web_app.py

Open http://localhost:8080 — point your camera at a LEGO part and tap capture.

CLI

python main.py

Uses assets/test.jpg by default (override with IMAGE_FILE env var).

Configuration

All settings are environment variables (see .env.example):

Variable	Default	Description
API_URL	https://api.brickognize.com/predict/	Brickognize endpoint
CONFIDENCE_THRESHOLD	0.5	Minimum detection confidence
REQUEST_TIMEOUT	5	API request timeout (seconds)
KMEANS_CLUSTERS	3	Color quantization clusters
MIN_VALID_PIXELS	50	Minimum pixels for color detection
CEREBRAS_API_KEY	—	LLM API key for sorting
CEREBRAS_MODEL	gpt-oss-120b	LLM model name
CEREBRAS_API_BASE	https://api.cerebras.ai/v1	LLM API base URL

Deploy on Render

1. Create a new Web Service on Render
2. Set Build Command to pip install -r requirements.txt
3. Set Start Command to python web_app.py
4. Add all env vars from .env.example in the Render dashboard
5. Deploy

Project structure

├── web_app.py         # FastAPI web server
├── main.py            # CLI entry point
├── color_detect.py    # Color detection via k-means
├── sort.py            # LLM-based part sorting
├── templates/         # Web frontend
└── data/              # Saved parts & photos