RSDiff
Copyright 2026 Ahmad Sebaq

This product is licensed under the Apache License, Version 2.0 (see LICENSE). The license
covers the code in this repository only.

Third-party components used at runtime, not distributed with RSDiff:

- imagen-pytorch (https://github.com/lucidrains/imagen-pytorch), MIT License. RSDiff's
  cascade is built from its Unet, Imagen and ImagenTrainer classes.
- T5-base text encoder and tokenizer (https://huggingface.co/google-t5/t5-base),
  Apache License 2.0, downloaded from the Hugging Face Hub on first use.
- OpenAI CLIP ViT-B/32 (https://huggingface.co/openai/clip-vit-base-patch32), MIT License,
  used only for evaluation.

The released checkpoint (https://huggingface.co/asebaq/rsdiff-sr-cascade-ep650) was
trained on the RSICD dataset (Lu et al., "Exploring Models and Data for Remote Sensing
Image Caption Generation", IEEE TGRS 2018). No RSICD images or captions are distributed
with this package. RSICD's terms for commercial use have not been established; see
docs/LICENSE_AUDIT.md before using the checkpoint or its outputs commercially.

Images produced by RSDiff are synthetic. They are not observations or predictions of any
real location.
