Lobe

Lobe is a desktop application originally developed to simplify the process of creating machine learning models for image recognition, classification, and object detection. It was designed to lower the barrier for users without a deep background in coding or data science, allowing them to train custom models using a visual, drag-and-drop interface. The primary goal of Lobe was to make machine learning accessible to creators, educators, hobbyists, and small businesses who wanted to integrate intelligent image analysis into their projects without writing complex algorithms. The application guided users through the entire workflow: collecting and labeling images, training a model on those examples, and then testing or exporting the trained model for use in other applications. Lobe supported image classification where users could define categories and provide example images for each category, and the tool would automatically learn to distinguish between them. It also offered object detection capabilities, enabling users to identify the location of objects within an image. Training was performed locally on the user's computer, leveraging built-in neural network architectures that had been optimized for speed and ease of use. After training, users could evaluate the model's performance with a simple test interface, seeing confidence scores for each prediction. The true value of Lobe lay in its export functionality: once a model was trained and validated, users could export it in several formats, including a TensorFlow model, an ONNX model, or a Core ML model for iOS. Additionally, Lobe provided starter kits for integrating the model into real-world applications.