SimpleML

来自 App Store 来源

EasyML streamlines the most tedious part of machine learning: preparing your training data. Built exclusively for macOS and optimized for Apple Silicon, it's the missing link between your photos and a trained Create ML object detection model. LABEL YOUR DATA Draw bounding boxes with precision using an intuitive canvas interface. Define custom labels, then work through your images efficiently. - Rapid cycle mode for repetitive labeling tasks - Smart "shy labels" automatically resolve overlapping boxes - Annotation versioning lets you maintain multiple labeling strategies - Native Create ML JSON export — drag straight into Create ML TRAIN, THEN LET AI HELP Here's where EasyML shines. After manually labeling a small batch: 1. Export and train a model in Create ML (Xcode Developer tool) 2. Import your .mlmodel back into EasyML 3. Let it auto-label your remaining images 4. Review predictions and correct mistakes 5. Retrain with your expanded dataset This iterative workflow can reduce labeling time by 80% or more. AUGMENT YOUR DATASET Small dataset? No problem. EasyML generates thousands of variations from your source images: - Geometric: rotation, flipping, scaling, translation, perspective, shear - Color: brightness, contrast, saturation, hue adjustment - Quality: gaussian noise, blur, sharpening, compression artifacts Smart anti-clipping technology detects your subject and ensures it stays in frame during transforms. Auto-detect padding matches background colors seamlessly. BUILT FOR YOUR MAC EasyML is a true native macOS application — no internet connection needed: - Metal GPU acceleration for blazing-fast batch processing - Optimized for Apple Silicon (M1/M2/M3/M4) - Dark mode support - All processing happens locally — your images never leave your Mac COMPLETE WORKFLOW EasyML handles: - Automatic train/validation splitting - Configurable output sizes (224×224 to 1024×1024) - Multiple export formats (PNG, JPEG, TIFF) - Progress tracking with time estimates - Preset augmentation profiles - Bounding box data set annotation - Custom ML assisted data labeling - CreateML ready export. PERFECT FOR - Students learning computer vision - Researchers building custom object detectors - Indie developers adding ML features to their apps - Hobbyists exploring machine learning - Anyone tired of enterprise pricing for basic tools

答案里的每个数字都来自本页上方展示的数据，可以逐条核对。