> ## Documentation Index
> Fetch the complete documentation index at: https://docs.worlds.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Collections & Models

> Index and tag video, generate training and test data, and train, evaluate, and deploy models.

Collections & Models is where video gets indexed and tagged, training and test data gets generated, and models get trained, evaluated, and deployed to production.

## Before You Start

* Available to Administrators and Super Administrators only. Because collections and models aren't tied to a specific site, and can pull together cameras from anywhere in your organization to run a shared detection model, this entire section is unavailable to Manager and User roles regardless of their individual site permissions.
* For step-by-step guidance on organizing collections, tagging, indexing, and training a model, see the [Model Training & Deployment Guide](/collections-and-models/model-training-deployment). This page covers where things live and how to track progress, not the training workflow itself.

\[SCREENSHOT: Collections & Models navigation]

## Collections

Collections are essentially folders for cameras. A camera can belong to multiple collections, and you use collections to group cameras that share similar environmental context so they can run the same detection model together, regardless of which site each camera physically belongs to.

\[SCREENSHOT: Collections list]

## My Workspace

A dashboard of every recent model-training-related job you've personally kicked off, training a model, generating data for a collection, indexing video, with quick links to jump straight into the output of each one.

Use this page to check on anything you've started recently and pick back up in the training pipeline once a job finishes.

\[SCREENSHOT: My Workspace - recent jobs list]

## Jobs

A full list of every internal system workflow running against any collection in your organization, not just the ones you started. Each entry shows its status, and clicking into one shows full details: task IDs and logs.

This page exists mainly as an internal tool for monitoring workflow timing and system status, but it's also useful if you're collaborating with others on tagging or training and need a quick way to check whether a job has finished before picking up the next piece of work.

\[SCREENSHOT: Jobs list with one job's detail view open]

## Active Processes

Any job in progress here also shows up as an icon in the Active Processes menu at the bottom left of the navigation, alongside your active detectors, with a live estimate of that job's percent completion.

## Quick Reference

* **Access** - Administrator and Super Administrator only; not available to Managers or Users regardless of site permissions
* **Collections** - folders that group cameras (from any site) sharing a detection model
* **My Workspace** - jobs you personally started
* **Jobs** - every job across the organization, plus task IDs and logs
* **Active Processes** - live status and completion percentage for any running job, visible from anywhere in the navigation
* **Deep dive** - see the [Model Training & Deployment Guide](/collections-and-models/model-training-deployment) for the actual training workflow