kby-ai/Fire-Smoke-Detection-Docker — reverse-engineered prompt

Build me a simple Dockerized fire and smoke detection service using the existing KBY AI SDK files in this repo. I want to be able to run it on Ubuntu with Docker, mount my license.txt file, and send an image from Postman or another app to see if there is fire or smoke in it.

Please make the app expose a POST endpoint for normal image upload at /fire and another one for a base64 encoded image at /fire_base64. The response should be easy to understand, with the detection result and any useful details the SDK returns. If the license is missing or invalid, show a clear message and print the machine code so I know what to send to KBY AI for a license.

Also include a very simple browser demo on port 9000 where I can upload a picture and see the result. Keep the setup easy, document the Docker build and run commands, and make sure the container starts cleanly.