Skip to content

Tutorials

This page points to practical tutorial material that exists in the repository today. Use Getting Started first if you have not yet verified that Annolid opens a video and saves annotations in your environment.

Choose a Tutorial Path

If you need to... Use
Learn the basic GUI tracking loop Tracking four interacting mice with one labeled frame and Workflows
Downsample or preprocess videos GUI video downsample workflow and Video Processing with FFmpeg
Score behavior events Behavior labeling with Timeline, Flags, and Annolid Bot
Define zones and export zone metrics Draw zones quickstart and Zone analysis workflow
Correct tracking drift or missing frames Segment-based batch tracking and Tracking correction with SAM3 Agent and Annolid Bot
Train or evaluate models Notebook tutorials in docs/tutorials/ and model-specific annolid-run help output
Run realtime camera workflows Realtime Wireless Camera Detection and Multi-Camera Realtime Detection

Video Walkthroughs

Annolid Track Animals dialog with model, video, checkpoint, and output controls

Focused Markdown Tutorials

Notebook Tutorials in docs/tutorials/

Representative notebooks currently tracked in the repo:

  • docs/tutorials/Extract_frames_from_a_video.ipynb
  • docs/tutorials/yolov8_tracking_tutorial.ipynb
  • docs/tutorials/Annolid_video_batch_inference.ipynb
  • docs/tutorials/Annolid_model_evaluation.ipynb
  • docs/tutorials/Annolid_Pose_Estimation_on_YOLO_Tutorial.ipynb
  • docs/tutorials/Annolid_Instance_Segmentation_on_YOLO_Tutorial.ipynb
  • docs/tutorials/YOLO_SAHI_inference_for_ultralytics.ipynb
  • docs/tutorials/zero_shot_object_detection_and_tracking_with_grounding_dino.ipynb
  • docs/tutorials/RAFT_optical_flow.ipynb
  • docs/tutorials/Annolid_behavior_video_classification_on_slowfast.ipynb

How To Choose

  • Start with the GUI/video workflow if you are labeling or reviewing data interactively.
  • Use the notebook tutorials when you need training, evaluation, or post-processing examples.
  • Use the markdown guides when you need an operational setup such as MCP, sandboxed shell execution, or batch tracking.
  • Use annolid-run help train <model> or annolid-run help predict <model> when a tutorial mentions a model backend and you need the current command-line options.

Repository-first Note

Some older docs still link to historical book content, but the current maintained tutorial sources for this repo live under docs/ and docs/tutorials/.