WEBVTT

00:00:00.300 --> 00:00:05.728
Image segmentation divides a photo into precise regions, pixel by pixel.

00:00:06.269 --> 00:00:12.870
In virtual try-on, it separates the person from the background, and the clothes from the skin and the hair.

00:00:13.530 --> 00:00:18.775
This map tells the AI exactly which pixels to change, and which pixels to leave alone.

00:00:19.698 --> 00:00:22.066
Pose estimation says how you stand.

00:00:22.552 --> 00:00:24.803
Segmentation says where everything is.

00:00:25.305 --> 00:00:27.340
That's image segmentation.