segmented

dev setup

deeplab

$ (install deeplab, publish-local)
$ yarn install
$ yarn link-local

Prep images

1. Get info

nodes = [];document.querySelectorAll('.collection-photo').forEach(n => nodes.push({src: n.src, alt: n.getAttribute('alt')})); console.log(JSON.stringify(nodes));

2. Fetch images (./fetch.sh)
3. Resize (eg cd public/img && mogrify -resize x224 -path ../img-resized *.*)

Prep masks

This builds the masks in the browser, writes them to disk with localforage, then forces downloading them to yield all the actual files. Then post-processing to resize and move into source for reading in production.

1. Clear localforage cache (?disabled&clear-localforage)
2. Run locally to build masks, with cache enabled, over the whole dataset (?disable-fetching-images)
3. Get the localforage dump, write it to cached.json
4. Run force download to get all the masks (?disable-fetching-images&force-download-for-dev), put them in the masks folder.
5. Resize (eg mkdir -p public/masks/v4-resized && cd public/masks/v4 && mogrify -resize x224 -path ../v4-resized *.*)
6. Verify it works.
7. Clear localforage cache (?disabled&clear-localforage)

This project was bootstrapped with Create React App.