Drone/ land camera image data and Machine-Learning based solution for predicting crop yield for optimal harvest planning.
Commercial agriculture/horticulture producers need to estimate crop yield in advance to help in optimal harvest planning. Absence of such information often results in ad-hoc planning of storage facilities, packaging materials, transportation, logistics, and labour. Over-provisioning results in higher costs while under-planning causes wastage of farm produce. With traditional farming expertise getting scarce, a reputed farm producer approached Accent for an appropriate digital solution.
Our field team made on-site visits to assess ground realities and challenges. Back in our Design Thinking Lab, a cross-sectional team churned out options to come up with a solution using precision imaging and Machine Learning. Continuous improvement of image data and fine-tuning of the machine learning model progressively enhanced prediction accuracy. Further refinements are ongoing with deployment of IoT-based sensors for soil, water and weather parameters. The algorithm can be further enhanced by correlating with historical data.
The Business Scenario:
- Farm produce company using contract farmers.
- Contract farms spread across geographies with no direct control on the farm operations.
- Needs advance estimation of crop yield for planning of the harvest season and post-harvest processing supply chain.
The Challenges Identified:
- Diminishing human expertise.
- Next generation not interested in the agri domain.
- Experts can’t visit all farms due to vast geographies and time constraints.
- Ad-hoc decisions cause financial burden or wastage of produce.
Technology-led Innovation:
- Images of early harvests using land-based and drone cameras.
- Machine learning (ML) models trained to identify the produce and estimate early harvest.
- Crop Yield Algorithm based on early harvest data and IoT inputs corelated with historical data.