Skip to main content
This guide walks you through the core end-to-end flow on the Repli5 platform—from uploading your first dataset and organizing team projects to choosing the right augmentation workflow, evaluating output quality, and exporting your finished training set.

​
End-to-End Platform Walkthrough

1

1. Gain Access & Set Your Workspace Focus

Apply for an account by emailing your address and use case to invite@repli5.com, then sign up using the invitation link sent to your inbox.On your first sign-in, the Repli5 platform asks a few quick questions about your domain and goals so it can tailor your workspace and open the workflow best suited to your problem. You can update these preferences at any time from your profile menu (Workspace focus).
2

2. Upload Your Data to the Data Library

Navigate to Data Library in the left sidebar to bring your source images into the Repli5 platform:
  • Upload a Dataset ZIP: Upload a .zip archive containing either a simple folder of RGB images (rgb/ or rgb_raw/) or a multi-modal dataset (with optional masks and bounding-box annotations).
  • Inspect, Tag & Filter: Open any dataset in the Data Library to browse individual datapoints at full resolution, create or assign tags, filter by tag, and inspect existing masks or bounding boxes.
  • Single-Image Quick Tests: If you want to experiment before uploading a full dataset, you can also upload a single image directly inside any workflow on the Augment page using the Upload tab.
3

3. Organize Your Work with Projects (or Work Standalone)

On the Repli5 platform, you can run augmentations either without a project (under All Projects in your personal workspace) or inside a dedicated Project shared with your team:
  • Create a Project: In the left sidebar under Projects, click + New, enter a project Name and optional Description, and click Create project.
  • Add or Remove Datasets: Click the ⋮ action menu next to any project in the sidebar and select Manage Datasets (or click Add to project on a dataset card in the Data Library) to attach datasets from your library to that project.
  • Invite Teammates & Manage Projects: From the same ⋮ project menu, you can Manage Users (invite teammates), Pin to Top, Rename / Edit Project, Duplicate Project, or Delete Project.
  • Automatic Project Scoping: Selecting a project in the sidebar scopes your Data Library, saved Templates, and Augment History to that project—and automatically opens the workflow that was last used in that project.
4

4. Choose the Right Augment Workflow for Your Application

Open Augment in the left sidebar. Augment is the core generation workspace of the Repli5 platform and offers specialized workflows depending on the computer-vision problem you are solving:

Defect Generation

Best for visual inspection with scarce defect data. Start from defect-free (OK) part images and synthesize realistic scratches, dents, tears, or wrinkles—complete with automatic pixel-accurate COCO segmentation masks and 2D bounding boxes.

Variation Generator

Best for expanding object & environment diversity. Take close-up photos of parts or products (with or without existing defects) and vary their color, surface finish, or lighting while preserving original geometry and surface wear.
5

5. Configure, Enhance & Launch Your Augmentation

Every workflow on the Repli5 platform follows a streamlined three-card layout:
  1. Card 1 — Input (Dataset / Upload): Pick a dataset (selected by default) and choose specific images or the entire dataset, or switch to the Upload tab to test on a single image.
  2. Card 2 — Templates & Prompts / Presets: Load a saved Template from your project or write a new prompt/preset. Click the Enhance button below any description box to let the AI expand a short note into a detailed, well-structured prompt, and save your configuration as a reusable template.
  3. Card 3 — Summary, Credits & Rubric: Check the total task count and Required credits (1 credit per generated image; click Request if you need a credit top-up), verify the Evaluation Rubric Set, and click Augment.
6

6. Review Quality Evaluations, Regenerate & Download

As your job runs, results stream into the Augment History panel at the bottom of the page:
  • Automated Quality Evaluation: Every generated image is scored automatically (1–5 scale) against workflow-specific metrics (see Quality Evaluation & Review). Images scoring below 3 on any metric are flagged for attention.
  • Compare, Approve & Regenerate: Open any image in the side-by-side comparator, mark it Approved or Rejected, or click Regenerate to re-run and replace an individual image in-place without re-running the whole batch.
  • Download from the Data Library: Every completed batch run is saved as a dataset in your Data Library. Open the Data Library and click Download to export the complete .zip archive (including rgb_out/, rgb_raw/, masks, and annotations_coco.json).

​
Profile & Workspace Settings

Click your Profile button at the bottom of the left sidebar to customize your experience on the Repli5 platform:
  • Interactive Workflow Guide: Launch a guided spotlight tour of the active workflow’s controls at any time.
  • Workspace focus: Re-open the onboarding survey to update your primary domain and workflow focus.
  • Change password: Update your account password.
  • Light Mode / Dark Mode: Toggle the interface between Dark Mode (default) and Light Mode.
  • Logout: Sign out of your current session.