Building Tomorrow’s Agriculture with IMX500 and Spresense at Sony AI AgTech Challenge

From May 2 to 4, UC Davis hosted a weekend of fast-paced problem-solving and creative energy as students came together for the first-ever Sony AI AgTech Challenge. The event, held in collaboration with the AI Institute for Next Generation Food Systems (AIFS), invited teams to tackle real agricultural challenges using the Raspberry Pi AI Camera with Sony’s IMX500 sensor and the Sony Spresense microcontroller board.

Over the course of three days, undergrads and grad students formed teams, brainstormed ideas, and built working prototypes. Their mission? To develop real-world solutions for the future of agriculture and food systems using Sony’s powerful technology. They received mentorship and guidance as they prototyped AI-powered solutions and presented their work in a final pitch session on Sunday, May 4.

The Final Projects

The WeedTrakr team developed a system to detect and map the presence of weeds in agricultural fields, supporting more efficient crop management. They integrated Sony’s Spresense main board with its GNSS module to capture real-time location data, while the IMX500 sensor handled on-device weed classification using edge AI. GNSS data was transmitted to a Raspberry Pi via serial communication, where it was paired with AI inference results. The combined output—location and classification—was then visualized on a custom-built React dashboard, allowing users to monitor weed presence spatially and in real time.

To support better harvest timing and crop evaluation, RipeRight used AI to analyze visual cues of tomato ripeness and estimate nutrient content. With models trained on image data, the team leveraged the Sony IMX500 sensor to detect signs of nutrients like iron, calcium, and potassium, bringing smart, in-field crop assessment one step closer to reality.

AutoDrive focused on autonomous navigation in agricultural settings. The team used the Sony Spresense main board to control a mobile robot capable of traversing farm environments. Spresense’s GNSS functionality enabled the robot to stay within predefined boundaries, ensuring safe and controlled movement. As the robot moved, it captured images and stored them locally on an SD card for later AI analysis—an approach that supports post-hoc insights without requiring constant connectivity.

AgChat combined edge AI and conversational interfaces to support smarter crop decisions. Using the Sony IMX500 sensor, the team classified fruits based on visible indicators of nutrient content, such as iron, calcium, and potassium levels. These results were then passed to an AI-powered chatbot, which offered suggestions on how to improve fruit quality, creating an interactive tool for farmers to get actionable insights from their crop data.