Welcome to annolid’s documentation!

Note

The modern, canonical docs are now published at: https://annolid.com/portal/

Annolid supports a video-first workflow: open a video, jump to informative frames, label, track, and export. Frame extraction is still available as an optional image-directory workflow when needed.

• Annolid User Guide
  • Overview
• Install Annolid
  • Requirements
  • Set up Conda environment
  • Install ffmpeg
  • Install Annolid
  • Windows: pycocotools
  • Linux: Qt “xcb” plugin error
  • Launch Annolid
  • Agent Safety Defaults and Rollback Operations
  • Install Detectron2 locally
  • Install Detectron2 on Google Colab
  • Optional: Install older version of Pytorch for YOLACT
• Read PDFs Aloud with AI Voice (Text-to-Speech)
  • Prerequisites
  • Open a PDF in Annolid
  • Option A: Speak a selection (fastest)
  • Option B: Click-to-read paragraphs (PDF.js reader mode)
  • Change voice, language, and speed
  • Troubleshooting
• Extract Frames
  • Extract desired number of frames from a video based on optical flow
  • Display optical flow while extracting frames with –show_flow=True
  • Save all the frames as images
  • Select frames randomly by reservoir sampling
  • Extract all the key frames from a video used by the compression methods
• Track animals and Auto labeling
• Output CSV format
• Cutie + DINO Body-Part Tracker
  • Key runtime parameters
  • Tuning guidance
• Video Depth Anything
  • Run in the GUI
  • Download checkpoints (optional)
  • Run from Python
  • Outputs
• Image Editing (Local Diffusion / GGUF)
  • Install (Diffusers)
  • CLI examples
  • GUI
• DINOv3 Keypoint Segmentation (DinoKPSEG)
  • Training
  • Inference (CLI)
  • Inference (GUI)
  • Training Wizard (GUI)
  • Evaluation
  • LabelMe to COCO (GUI)
  • COCO to LabelMe (GUI)
  • Running Inference (GUI)
• Agent Tools Developer Guide
  • Operational Model
  • How to add a tool
  • Artifact layout
  • Caching semantics
  • Citation management tools
  • Operator Commands
  • Shell Session Tools
  • Improvement Quality Loop
  • Governance and Audit
  • Three.js bot tools
  • Browser Automation Safety
  • Annolid code/docs Q&A and tutorials
  • Realtime camera snapshot + email
  • Security and policy hardening (Phase 2)
  • Session memory and replay
• Annolid Bot Provider and Model Setup Tutorial
  • 1. Where Annolid reads settings
  • 2. Core environment variables
  • 3. Configure local models (Ollama)
  • 4. Configure OpenAI models
  • 5. Configure OpenRouter (OpenAI-compatible path)
  • 6. Configure Gemini
  • 7. Configure agent profiles and model routing
  • 8. Configure search key for tool calls
  • 9. Validate your setup
  • 10. Recommended secure workflow
• Annolid Bot Background Services Setup
  • 1. Install optional extras you actually need
  • 2. Minimal safe config (~/.annolid/agent/config.json)
  • 3. Enable WhatsApp (optional)
  • 4. Enable Zulip (optional)
  • 5. Enable Google Calendar (optional)
  • 6. Common errors and quick fixes
  • 7. Verify from your environment
• Annolid Bot Google Calendar Tutorial
  • 1. Install optional dependency
  • 2. Create Google OAuth desktop credentials
  • 3. Configure Annolid
  • 4. Example prompts
  • 5. Security notes
• Annolid Bot ClawHub Skills Tutorial
  • 1. Prerequisites
  • 2. Open Annolid Bot
  • 3. Search for skills (natural language)
  • 4. Install a skill by slug
  • 5. Restart behavior
  • 6. Troubleshooting
  • 7. What tools are used
• Annolid Bot WhatsApp Setup Tutorial
  • 1. Default: Embedded QR bridge mode
  • 2. QR flow
  • 3. No-reply troubleshooting (QR bridge)
  • 4. Optional: Cloud API mode
  • 5. Optional: External bridge mode
  • 6. Security checklist
• LLM API Key Security
  • What Annolid Stores
  • Recommended Key Setup
  • File Permission Hardening
  • Safe Operational Practices
  • Verify Configuration
  • Adding New Providers in GUI
• YOLOE-26 Prompting (Text / Visual / Prompt-free)
  • CLI (recommended for reproducible runs)
  • GUI (video inference)
• Config keypoint connection rules, events, and instances
• FAQs
  • Threshold based object segmentation
  • Convert WMV format to mp4 format using ffmpeg
  • Save the extracted frames to a user selected output directory
  • How to track multiple objects in the video?
  • How to convert labelme labeled dataset to COCO format?
  • Convert the tracking results csv file to Glitter2 csv format
  • Convert the keypoint annotations to labelme format

Indices and tables

• Index

• Module Index

• Search Page