Scalable Forecasting on Hydroponics Data
BUSINESS PROBLEM
- A hydroponic experiment used sensor- and camera-equipped units to grow various plants under uniform conditions, capturing continuous environmental and growth data.
- The challenge was to leverage sensor and camera data to forecast optimal plant growth and enable remote, data-driven decision-making for managing hydroponic systems more efficiently.
SOLUTION
- Conducted controlled plant experiments with various plants in a standardized hydroponic environment to ensure consistent data.
- Manipulated & converted RGB and IR images to quantify and monitor plant growth.
- The integrated data pipeline aligned high-frequency sensor streams with processed image data for analytics.
- Further, Machine learning models predicted plant growth 24 hours in advance, enabling proactive monitoring.
FORECASTING @SCALE WITH IMAGE & SENSOR DATA
BENEFITS
- Helps better understand the requirements of growing environment components.
- Enabling machine learning–based remote monitoring with ability to scale and adaptive to new scenarios.
CHALLENGES
Measure the Growth of plants using time-stamped RGB and IR images attached to the unit.
Formulating machine learning problem aligned with unique business problem with image and sensor data.
PERFORMANCE
Approx. 20% error rate in plant growth forecasts.
