This app expands on an app that uses semantic segmentation to segment out a person from background noise in a video stream and replace the background with an image or blur it out. This app demonstrates how to subsequently smooth segmentation boundaries. This app also demonstrates how to separate your app configuration information into a separate JSON file. For more details on this aspect of the app, please see the original blog post.
This app uses a config.json file to specify different options:
- model_id: specifies the alwaysAI model you'd like to use. Find more details on the available semantic segmentation models here
- target_labels: use this variable to pull out desired objects. This will depend on the model you use. While using "alwaysai/fcn_alexnet_pascal_voc", for instance, and wanting to segment out people, the label is 'person'. Add labels you wish to segment out in this list.
- background_images: the folder that stores your possible backgrounds
- image: the name of the image file to use for the background
- blur: whether to blur the background (
trueto blur,falseto not) -blur_level: a higher number will blur the background more - use_background_image:
trueindicates that the background should be replaced with the value set forimage
See this page for details on configuring projects with alwaysAI. Please see this document for details on building and deploying applications.
You should see an image such as the image below when using the 'beach.png' with 'use_background_image' and no blur in the config.json file.
Docs: https://dashboard.alwaysai.co/docs/getting_started/introduction.html
Community Discord: https://discord.gg/alwaysai
Email: support@alwaysai.co