peekaboo Peekaboo: Text to Image Diffusion Models are Zero-Shot Segmentors Here's a Google Colab Link An older Colab link