TerraSearch
OCR und Suche in digitalisierten Bodenuntersuchungsdokumenten
Tech Stack
| Part | Technology |
|---|---|
| Frontend | Flask (Web), Kibana (DB) |
| Backend | Elasticsearch (Storage, Indexing) |
| OCR | Google Tesseract, jTessBoxEditor (Training) |
| NLP | spaCy, ContextualSpellCheck, LocationTagger |
| Deployment | Docker/docker-compose (Containerisation) |
Setup and Installation
- Install Dependencies (Docker, docker-compose)
- Create .env File (see example below)
- Create Docker Network for Project
docker network create terrasearch-net - Start Docker Compose
docker-compose up -d
Environment File (.env) Example
VERSION="0.5.3"
# Username for elasticsearch
ELASTIC_USER="elastic"
# Password for the 'elastic' user (at least 6 characters)
ELASTIC_PASSWORD="elastic123"
# API Key for Google Maps
GOOGLE_API_KEY="GetMeFromGoogleCloudPlatform"
# Password for the 'kibana_system' user (at least 6 characters)
KIBANA_PASSWORD="kibana123"
# Version of Elastic products
STACK_VERSION=8.3.1
# Set to 'basic' or 'trial' to automatically start the 30-day trial
LICENSE=basic
# Port to expose Elasticsearch HTTP API to the host
ES_PORT=9200
# Port to expose Kibana to the host
KIBANA_PORT=5001
# Port to expose Frontend to the host
FLASK_PORT=8443
# Increase or decrease based on the available host memory (in bytes)
MEM_LIMIT=4294967296
# Set the cluster name
CLUSTER_NAME=terrasearch-cluster
# Sync directory inside container
SYNC_DIR="/data/sync"
# Local Dir containing pdf files
SYNC_DIR_LOCAL="C:\Downloads"