Dataset Manager
Each dataset is a folder with originals/ and outputs/. Run opens the pipeline and results.
Listings
…
Images
…
Dataset folders
…
Dataset folders
On-disk library · originals + outputs
Loading folders…
Upload & Auto Process Dataset
Upload a product-feed TSV/CSV. Every listing and image URL is imported and downloaded, then one click runs the full pipeline (analyze → route → mask → generate → QA) automatically.
Drop a product-feed TSV/CSV here (e.g. google_product_feed.tsv)
parsed via /datasets/auto-import — duplicate image columns are expanded
Create dataset
Upload
Drop one JPG/PNG here
uploaded via /images/upload
All categories
Named folders above are the main workflow. This table also lists leftover source categories.
Loading…
Images by state
No state data
Populated once images move through the pipeline.