FieldDay: No-Code Vision AI Training on Your Phone

FieldDay is a mobile ML toolkit that puts the power of custom computer vision into the hands of everyday users, no coding required. Instead of relying on generic pre-trained models, you can capture your own datasets using your phone's camera, iteratively train recognition algorithms, and correct their mistakes on the fly to sharpen accuracy for your specific use case. Whether you're identifying plant species, sorting recyclables, or spotting a particular tool in a workshop, FieldDay adapts to whatever visual task you throw at it. Beyond training, FieldDay is built to plug into the rest of your digital life. Models can trigger HomeKit automations, so your camera can flip on lights or control outlets based on what it sees, or be wired into Siri Shortcuts for hands-free, AI-powered actions. When you're ready to scale beyond the app, you can export trained models to other platforms and formats, making FieldDay a flexible starting point for hobbyist projects, classroom experiments, and even early-stage app prototypes.

1. Open FieldDay and start a new project to define what you want your vision AI to recognize. 2. Use your phone's camera to collect a custom dataset of images representing your target objects or scenes. 3. Train the model directly on your device and test its predictions in real time. 4. Correct any misidentifications to iteratively fine-tune accuracy. 5. Connect the trained model to HomeKit devices to trigger automations based on camera input. 6. Set up Siri Shortcuts to run AI predictions hands-free. 7. Export your finished model to other platforms or formats to use it in broader projects.