Wildlife Watcher Data Tools

Tools and scripts for manipulating the wildlife AI camera data to analyse the footage, create your own models and transfer them to the cameras.

This repository contains good tools for formatting data for the Weta Watcher project. Typically you would want to use this package if you:

• Want to create datasets from videos captured using camera traps
• Annotate your dataset
• Export for training in Edge Impulse

Installing

Clone this repo and install it using

pip install <path-to-repo>

Usage

You can either use the actions using python functions in scripts or notebooks or run actions using our CLI.

Actions

Supported actions:

• filter_empty_videos: Removes videos where no movement occurs.
• create_dataset: Creates a dataset in FiftyOne.
• show_dataset: Launch FiftyOne App where dataset can be inspected.
• list_datasets: List your datasets.
• create_annotation_job: Send dataset to CVAT for annotation.
• read_annotations: Read annotations from CVAT back to FiftyOne.
• preprocess_dataset: Process dataset to given FPS and size.
• export_dataset: Export dataset to disk in either a FiftyOne format or Edge Impulse.
• delete_dataset: Delete dataset from FiftyOne.

All actions can be found at src/wai_data_tools_actions.py

CLI

To check out the CLI install the package and run the following in your terminal to get started:

wildlifeai-cli --help

Note

This project has been set up using PyScaffold 4.2.1. For details and usage information on PyScaffold see https://pyscaffold.org/.