# Source and upstream notices

The project owner's original code and trained weights are released under the
root MIT LICENSE. This does not relicense the separately attributed third-party
panel driver, ESP-IDF components, or training images.

The trained INT8 flower model was supplied by the project owner from their
existing ESP32-S3 flower-GAN project. Its architecture and precision are preserved.
That project's handoff identifies Oxford Flowers102 (8,189 images, 102 classes)
as its training data: https://www.robots.ox.ac.uk/~vgg/data/flowers/102/ .
No training images or original badge data are included here. This notice does not
replace upstream model/dataset terms.

The modified panel driver in `source/firmware/main/epaper.c` adapts Gordon Williams /
Pur3 Ltd's Espruino ePaper Badge driver, under Mozilla Public License 2.0. Its source
is included in this release and the license is at `source/licenses/MPL-2.0.txt`.
Original: https://github.com/espruino/Espruino/blob/master/libs/js/misc/epaper_badge.js

The native application and bootloader use Espressif ESP-IDF v5.5.1. IDF and its
bundled component notices are retained in `source/licenses/esp-idf/`.
https://github.com/espressif/esp-idf/tree/v5.5.1

Flashing instructions use Espressif's official browser utility (not bundled):
https://github.com/espressif/esptool-js . Optional terminal instructions use
esptool 5.6.0, installed separately from PyPI under its upstream terms:
https://github.com/espressif/esptool . No custom installer is included.