streamware

Track Person Example

Fast person tracking using YOLO without LLM.

Quick Start

# Natural language
sq watch "track person"

# With custom URL
sq watch "track person" --url rtsp://camera/stream

# Polish
sq watch "śledź osoby"

Features

• Fast detection: ~1 FPS using YOLO only
• Movement tracking: Position, direction, entering/exiting
• TTS announcements: Voice alerts on changes

Output Messages

Event	Message
Person appears	“Person entering from left”
Position	“Person on left detected”
Movement	“Person moving right”
Exit	“Person left to the right”

Configuration

# .env settings for track mode
SQ_STREAM_FPS=1.0
SQ_YOLO_SKIP_LLM_THRESHOLD=0.3
SQ_USE_GUARDER=false

Python API

from streamware.intent import parse_intent, apply_intent
from streamware.core import flow

# Parse natural language
intent = parse_intent("track person")
apply_intent(intent)

# Run
result = flow(f"live://narrator?source={url}&mode=track&focus=person").run()

This site is open source. Improve this page.