Intro

Here you can find the source code of the telegram bot that interacts with the inference server that fetches the model from the roboflow server

Prerequisites

1. 12 Gb space for docker images
2. Start Docker inference server on localhost:9001. You can do it with: docker-compose.yaml
3. Set up environment variables for Java:
   1. API_KEY: Roboflow private key from app.roboflow.com/<your_name>/settings/api
   2. BOT_TOKEN: Telegram bot API token retrieved from BotFather
   3. OWNER_ID: Your telegram id or anyone who will receive feedback and other admin information from users

How to

Deploy

1. You can build the image of the bot from the Dockerfile using: docker build . -t my-bot:latest
2. Then you should replace services.telegram-bot.image value to my-bot:latest
3. And run everything with docker-compose up

Use

Send a picture (so-called photo) to the bot. Photo is the image with compression.

Experiment

1. Tune confidence and iou-threshold in application.yaml
2. Change model-id to yours. You can find it on Model page of your dataset on Roboflow.

Example of interaction