Components

Segment Image (Ultralytics YOLO)

# Segment Image (Ultralytics Yolo) Summary: Instance segmentati

Classify Image (Ultralytics YOLO)

# Classify Image (Ultralytics Yolo) Summary: Image classificati

Input Image File

# Stream images from files with controllable timing Use This Wh

Check Object Distance

# Monitor proximity and trigger distance-based alerts Use This

Track Bounding Boxes

# Track Bounding Boxes Summary: Multi‑frame association that as

Detect Zone Transition

# Detect zone entry/exit events for tracked objects Use This Wh

Detect Fallen People

# Detect people who have fallen Use This When - Building safety

Input Video File

# Decode and stream video frames from file Use This When - Test

Detect Objects (Ultralytics YOLO)

# Detect Objects (Ultralytics Yolo) Use This When - Building re

Evaluate Expression

# Evaluate Python expressions on pipeline data Use This When -

Send Object Count MQTT

# Send Counts (MQTT) Summary: Publishes total and per‑class obj

Visualize Expression

# Visualize Expression General-purpose image transformation/eff

Query VLLM MiniCPM-Llama

Initial Commit

Track Objects (Ultralytics YOLO)

# Track Objects (Ultralytics Yolo) Summary: Runs Ultralytics tr

Input Text File

# Input Text File Use This When - Reading one or more text file

Reidentify Objects

# Reidentify Objects Summary: Assigns stable global IDs by buil

Calculate Optical Flow (RAFT, Triton)

# Compute dense optical flow between frames Use This When - Tra

Classify Image (Triton)

# Classify Image Summary: Generic image classification that out

Visualize Zone Transition

# Visualize Zone Transition Summary: Draws a polygonal zone and

Detect Landmarks (Triton)

# Detect pose landmarks within bounding boxes Use This When - B

Send MQTT Cumulocity

# Send PPE Status (MQTT) Summary: Publishes per‑person PPE stat

Detect Objects (Triton)

# Detect objects using Triton-served models Use This When - Bui

Calculate Kalman Filter Position (2D)

# Smooth tracked positions with Kalman filter Use This When - T

Estimate Crowd Count

# Estimate crowd count using density map prediction Use This Wh

Estimate Depth (HF)

# Estimate monocular depth using HuggingFace pipeline Use This

Estimate Density Map Peaks

# Extract individual person locations from crowd density maps U

Calculate Object Movement

# Extract per-object motion vectors from optical flow Use This

Input Browser Webcam

# Capture webcam and microphone from browser via WebRTC Use Thi

Input PDF

# Extract text from PDF documents for LLM analysis Use This Whe

Label Animal-AP-10K Landmarks

# Convert landmark list to AP-10K animal 17-keypoint schema Use

Label Halpe-Body Landmarks

# Convert landmark list to Halpe 26-keypoint full body schema U

Limit Message Rate

# Throttle message stream to target rate by selective forwarding

Log Message

# Print messages to logs for debugging and inspection Use This

Output Browser Video

# Stream video and audio to browser via WebRTC with low latency

Transcribe Audio (Openwhisper)

# Transcribe Audio Summary: Local speech‑to‑text that supports

Caption Image (BLIP)

# Generate natural language image captions Use This When - Buil

Detect Voice Activity (Silero Vad)

# Detect voice activity using Silero VAD Use This When - Buildi

Visualize Vector Field

# Visualize Vector Field Summary: Converts dense optical flow t

Animate Face

# Animate face Use This When - Building virtual avatar systems

Visualize

# Visualize Summary: Fast overlays for boxes, masks, landmarks,

Locate Bodypart

# Extract body part regions from pose landmarks Use This When -

Denoise Audio (MP-SENet)

# Remove noise from audio streams Use This When - Preprocessing

Generate Speech (Kokoro)

# Generate speech using Kokoro TTS models Use This When - Build

Collect

# Accumulate stream items until triggered to flush Use This Whe

Detect Object Containment

# Test if bounding box anchor falls within polygon ROI Use This

Detect Objects (HF)

# Detect objects using HuggingFace models with automatic backend

Detect Turn Speech

# Detect speech turn completion for conversation flow control U

Input Video URL

# Ingest live video streams from network URLs Use This When - B

Label COCO-Body Landmarks

# Convert landmark list to COCO 17-keypoint body schema Use Thi

Label COCO-MPII-Body Landmarks

# Convert landmarks to Human22 schema with COCO or Halpe mode U

Label COCO-Wholebody-Hand Landmarks

# Convert landmark list to Hand21 schema for hand gesture analys

Project Plane

# Transform 2D points using homography and lens distortion corre

Query AI (Ollama)

# Query Ollama LLM with streaming or batch responses Use This W

Query Vision (Ollama)

# Query vision LLM via Ollama Use This When - Running vision-la

Summarize Tracked PPE

# Summarize Tracked PPE Summary: Computes PPE compliance over t

Segment Masks (Triton)

# Segment Masks Summary: Pixel‑accurate segmentation that emits

Query OCR (Surya)

# Query OCR Use This When - You want one OCR worker with plugga

Transcribe Audio (SenseVoice)

# Transcribe Audio (SenseVoice) Summary: CTC-based speech-to-te

Transcribe Audio (Moonshine)

# Transcribe Audio (Moonshine) Summary: Ultra-lightweight speec

Transcribe Audio (Faster Whisper)

# Transcribe Audio (Faster Whisper) Summary: CTranslate2-accele

Transcribe Audio (Parakeet)

# Transcribe Audio (Parakeet) Summary: High-accuracy speech-to-

Filter Bbox Pose

# Filter Emits the second input only when the first input (a bo