2_app_tour.md

from requests import sessionfrom requests import session

Walk through for the application

There will be limited code during this section so we will just use the terminal

python3

import fiftyone as fo

fo.list_datasets()

dataset = fo.load_dataset("quickstart")

session = fo.launch_app(dataset)

#If we ran this in code we would need to hold the session open to prevent the app server from exiting
# session.wait()

# Codespaces doesn't play well with websockets or SSE so at the end of our URLs we need to put
# ?polling=true
# Which adds long polling

Things to do

1. Look at the datasets available
2. Left side and what you can do with the fields
3. selecting images - show how that appears in the API and vice versa https://docs.voxel51.com/api/fiftyone.core.session.html?highlight=session#fiftyone.core.session.Session

selected_samples = session.selected

session.select_samples(["677e09e04b4ee07986bd0c79"])

4. looking at an individual image
5. Tag labels or samples in the app
6. Lightning = indexed field. To turn off toggle query performance mode
7. Filter gear
8. Change color schemes
9. Plugins

https://docs.voxel51.com/plugins/index.html

https://github.com/voxel51/fiftyone-plugins

Plugin plugin

fiftyone plugins download https://github.com/voxel51/fiftyone-plugins --plugin-names @voxel51/plugins

Image Quality data quality - There is a teams feature that does this and more https://docs.voxel51.com/teams/data_quality.html

fiftyone plugins download https://github.com/jacobmarks/image-quality-issues/

refresh the page then [ctrl] + `, then "common issues". Let's do brightness

Build a dashboard https://github.com/voxel51/fiftyone-plugins/tree/main/plugins/dashboard

Groundtruth histogram and then scatterplot of Uniqueness versus Brightness

fiftyone plugins download https://github.com/voxel51/fiftyone-plugins     --plugin-names @voxel51/dashboard

8. Finally, you can change the data in the view from the API.

session.dataset = dataset.limit(5).clone('little_data')