peekaboo

Peekaboo: Text to Image Diffusion Models are Zero-Shot Segmentors

Here's a Google Colab Link

An older Colab link