Overview:

• Early algorithm validation has shown identification capabilities down to the individual plant, far exceeding expectations.

• Ability to produce prescription maps for integration with variable rate technology has been successfully validated.

In August 2018, Challenge Advisory collaborated with one of the world’s largest food processors to implement a technology solution to detect, classify and geo-locate specific anomalies, pests and diseases which were directly impacting the bottom line of their farmers.

The food processor is currently one of the world’s largest producers of frozen potatoes, working with over 2,750 farms globally and owning 24 processing plants across North America, Europe and Asia. The integrity of their vertical supply chains is an essential component of the firm’s competitive advantage, yet 89% of their growers had reported experiencing a pest or disease infestation with cost implications between $146-$700 per hectare within the last 18 months.

Tasks:

• Assess the specific pests, diseases and anomalies which had the greatest impact on potato growers for the processor.
• Utilise all existing data sources for early identification of problems and identify relevant new required data sources.
• Data interpretation must provide a recommended course of action for growers.
• Long term goal is integration with variable rate technology.

Strategy:

• Identify Problem – Solution Fit: Step 1 was to collaborate with farmers to identify potato specific pests, infestations and diseases which had the greatest impact on profitability.
• Data Pooling: Step 2 was to assess the data sources available to farmers, and which alternative data sources would help us with early identification.
• Training AI Model: Step 3 was to work with our industry partner to train their AI model with the sample data provided by the processor and farms.
• Actionable Recommendations: Step 4 was to ensure actionable recommendations were provided by our technology partner to provide farmers with deep insights for early detection.
• Build Prescriptions for VRT: Step 5 required producing prescription maps for integration with existing variable rate technology deployed on the processors farms.

Outcome:

Our Agricultural Analysts were deployed to the processors Supply Chain Operations team and subsequently to 4 pilot farms to begin identifying the largest pests, disease and infestations faced by growers. The analysts also identified unique challenges technology solutions must overcome, to be adopted at the farm level.

Analysts Identifying Problem:

• Colorado Beetle: Our agricultural analysts found the pest had cost implications of $302 per acre and could impact up to 29% of US farm supplier and 21% of European farms supplying to the client.
• Late Blight: This disease had the highest cost implication at over $493 per hectare not including non-fungicide control costs. However, the disease was disproportionately affecting farm suppliers from Idaho and Washington, and Poland.
• Virus Y: Our analysts found Virus Y had cost implications of $4million per year across the clients suppliers from Europe and North America.

Data Pooling:

After identifying the problem, Challenge Advisory collaborated with 4 pilot farms, local agronomists and the food processor to begin collating critical data. This included Satellite, Drones (Fixed & Quad Wing), Machine Vision Cameras (Including tractor mounted cameras), Field Sensors and Yield Monitoring.

Impact of Solution:

Working with our industry and technology Partners, the solution is currently deployed on a pilot project focused on early detection of the Colorado Beetle and Late Blight for potato production.

• Early algorithm validation has shown identification capabilities down to the individual plant, far exceeding expectations.
• POC showed identification and detection capabilities of Colorado Beetle reduced the cost implications from £302 per hectare to $221.
• Algorithm validated for earlier detection of Late Blight averaging 18% reduction in fungicide use and 6% savings in storage costs.
• Ability to produce prescription maps for integration with variable rate technology has been successfully validated.