| Service | Status |
|---|---|
| Backend-API |  |
| ESP32-Cam Firmware |  |
This project was developed as part of the course “Selected Topics in Data Science and AI in Business” at DHBW Ravensburg. It focuses on building a low-cost vision system using an ESP32-CAM, which captures images and sends them to a server running a segmentation model. The model detects and classifies simple geometric shapes (black circles) for use in quality assurance and component sorting applications.
See ESP Readme for all instructions regarding the ESP32 including setup, updating the firmware and developer instructions.
Start containers using the published image:
docker compose up -d➡️ Uses the prebuilt image paulgrbr/hivehive-be:latest from Docker Hub.
Build image locally and start the dev environment:
docker compose -f docker-compose-dev.yml up --build➡️ Builds the backend image from backend-api/Dockerfile and runs the container.
The backend uses an AWS-compatible S3 storage endpoint to store all raw incoming images. These images can be used for training, validation, and dataset management.
➡️ If left out, the backend will skip this step. Configure these values in a .env file or via Docker Compose. A .env.template file is provided.
AWS_ENDPOINT="https://[your-endpoint-here]"
# Access credentials
AWS_ACCESS_KEY_ID="[your-access-key-id-here]"
AWS_SECRET_ACCESS_KEY="[your-secret-access-key-here]"