Temporal Event Detection
Sliding window search + temporal aggregation + threshold alerts. This is just retrieval + time—not a special system.
"Alert when violence or weapons appear in live stream footage from the last 5 minutes with 85% confidence"
Why This Matters
Temporal queries are retrieval operations with time constraints. Frame-level embeddings enable precise event detection.
from mixpeek import Mixpeekclient = Mixpeek(api_key="your-api-key")# Create alert condition retrieveralert_retriever = client.retrievers.create(retriever_name="safety_monitor",stages=[{"stage_id": "feature_search","config": {"query_text": "violence OR weapons OR threatening behavior"}},{"stage_id": "score_filter","config": {"min_score": 0.85}}],webhook_url="https://api.company.com/alerts")# Monitor live streamstream_results = client.retrievers.execute(retriever_id=alert_retriever.id,inputs={"collection_id": "live_streams","time_window": "last_5_minutes"})# Trigger alert if matches foundif len(stream_results['documents']) > 0:client.webhooks.trigger(webhook_url=alert_retriever.webhook_url,payload=stream_results)
Retrieval Flow
Semantic match against alert conditions
Time window filtering
Confidence threshold
Feature Extractors
Feature Extractors
Video Embedding
Generate vector embeddings for video content
Audio Transcription
Transcribe audio content to text
Text Embedding
Extract semantic embeddings from documents, transcripts and text content
Retriever Stages
feature search
Search collections using multimodal embeddings
attribute filter
Filter documents by metadata attributes
score filter
Filter documents by relevance score threshold
