TitikPanen

Dashboard
List Lahan
Detail Lahan
Riwayat

Inspiration

Indonesia is one of the world's largest chili producers, yet farmers still face uncertainty when deciding the right time to harvest. Harvesting too early can reduce quality and selling price, while harvesting too late increases the risk of spoilage, disease, and losses due to unexpected weather. Most existing agricultural technologies focus on monitoring environmental conditions but rarely provide actionable recommendations.

We wanted to build a system that goes beyond monitoring. Instead of simply showing sensor values or camera images, we asked a simple question:

"What should farmers do today?"

This idea led us to develop TitikPanen, a Regional Harvest Intelligence Platform that combines Artificial Intelligence, IoT, weather intelligence, and an expert decision engine to support smarter harvest planning.

What it does

TitikPanen is a web-based Decision Support System that helps farmers, cooperatives, and food authorities determine harvest priorities using AI-powered crop analysis and weather information. The platform captures images of chili plants using a portable ESP32-CAM node. The images are analyzed by a Computer Vision model to estimate fruit maturity, visible fruit conditions, and plant health. Weather information is then retrieved from BMKG and combined with AI observations through our Harvest Intelligence Engine (HIE).

Instead of only displaying raw AI predictions, HIE transforms them into practical recommendations such as:

Harvest Now
Continue Monitoring
Harvest Earlier Due to Rain Forecast
High Disease Risk
Field Inspection Recommended

Different stakeholders receive different perspectives:

Farmers receive recommendations for their own fields.
Cooperatives can prioritize harvests across member farms.
Food Authorities gain regional insights to support production planning and food supply management. The platform is fully responsive and accessible from both desktop and mobile devices.

How we built it

We divided the system into four major components:

1. AI Computer Vision

A computer vision model analyzes uploaded chili images to estimate:

Fruit maturity
Fruit visibility
Plant condition
Confidence score

2. Harvest Intelligence Engine (HIE)

Instead of allowing AI to directly decide harvest timing, we developed a rule-based expert system that combines:

AI observations
Weather forecasts
Agricultural knowledge This produces explainable and actionable recommendations.

3. Backend

The backend was developed using FastAPI and acts as the brain of the platform by:

Receiving uploaded images
Connecting with the AI inference service
Retrieving weather data
Executing HIE rules
Storing prediction history
Serving APIs to the frontend

4. Frontend

The frontend was built as a responsive web application, providing:

Regional harvest map
Harvest readiness dashboard
Recommendation panel
Weather alerts
Farm analytics
Mobile-friendly interface

Challenges we ran into

One of the biggest challenges was realizing that predicting an exact harvest date is not scientifically reliable for chili plants. Chili is harvested multiple times during its lifecycle, and fruit maturity varies significantly even within the same plant. Another challenge was image occlusion. A single camera cannot observe every fruit because many are hidden behind leaves or branches. Instead of attempting absolute fruit counting, we redesigned the system to estimate visible harvest readiness and use it as one component of a larger decision-making process. Time constraints also pushed us to rethink our architecture. Rather than building an overly complex predictive AI model, we focused on creating an explainable hybrid system where AI performs observations while the Harvest Intelligence Engine generates recommendations.

Accomplishments that we're proud of

Our biggest achievement was shifting the project from a simple crop monitoring application into a complete Decision Support System. Instead of stopping at image detection, we successfully integrated:

AI Computer Vision
Weather Intelligence
Rule-Based Expert System
Regional Dashboard
Multi-role Decision Support We also designed the platform to serve multiple stakeholders, allowing the same data to generate different insights for farmers, cooperatives, and food authorities. Most importantly, every recommendation generated by the system is explainable, making it easier for users to trust and understand the decision process.

What we learned

Throughout this project we learned that AI alone is not enough for real-world agriculture. Computer Vision is excellent at observing plant conditions, but agricultural decision-making also requires contextual information such as weather forecasts, cultivation practices, and expert knowledge. We also learned the importance of building explainable AI systems. Farmers need understandable recommendations rather than black-box predictions. Combining AI with a rule-based expert system creates a more transparent and practical solution. Finally, this project reinforced the value of interdisciplinary collaboration between AI, IoT, software engineering, and agricultural knowledge.

What's next for TitikPanen

Our long-term vision is to transform TitikPanen into a comprehensive Regional Harvest Intelligence Platform that not only supports harvest decisions but also orchestrates the entire agricultural supply chain. The next phase of development focuses on building an AI-powered Supply Chain Management (SCM) ecosystem, where harvest planning, distribution, logistics, and market demand are intelligently connected. By combining regional harvest insights with production forecasts and demand analysis, the platform will be able to automatically recommend optimal harvesting schedules, distribution routes, warehouse allocation, and delivery priorities to reduce post-harvest losses and improve supply chain efficiency. On the AI side, we plan to develop more accurate computer vision models using larger real-world agricultural datasets, multi-view image acquisition, and additional environmental data from IoT microclimate sensors. These improvements will increase harvest readiness estimation accuracy, disease detection performance, and yield prediction capabilities. In the future, TitikPanen aims to become an end-to-end agricultural intelligence platform that empowers farmers, cooperatives, distributors, and food authorities with data-driven insights from field monitoring and harvest decision support to intelligent supply chain and distribution management.