v1.2.2

• Fixed iOS/Android Consistency issues
• Added Platform Considerations section explaining expected cross-platform variances
• Fixed broken Table of Contents hyperlinks in README
• Misc bug fixes

v1.2.0

This release adds powerful new data augmentation capabilities for ML training pipelines.

✨ New Features

Color Jitter (colorJitter)

Granular color augmentation with fine-grained control over individual color properties:

• Brightness - Additive adjustment with symmetric or asymmetric ranges
• Contrast - Multiplicative adjustment around mean
• Saturation - Multiplicative adjustment (0 = grayscale)
• Hue - Cyclic shift as fraction of color wheel

Key capabilities:

• 🎯 Range support - Single value for symmetric range, tuple [min, max] for asymmetric
• 🔁 Reproducible - Optional seed parameter for deterministic augmentation
• 📊 Transparency - Returns all applied values for logging/debugging

v1.0.0 - Initial Release

A high-performance React Native library for image preprocessing optimized for ML/AI inference pipelines.

✨ Features

Core Image Processing

• 🚀 Native implementations in Swift (iOS) and Kotlin (Android)
• 🎨 Multiple color formats: RGB, RGBA, BGR, BGRA, Grayscale, HSV, HSL, LAB, YUV, YCbCr
• 📐 Flexible resizing: cover, contain, stretch, letterbox strategies
• 🔢 ML-ready normalization: ImageNet, TensorFlow, custom presets
• 📊 Multiple data layouts: HWC, CHW, NHWC, NCHW (PyTorch/TensorFlow compatible)

ML Pipeline Support

• 🤖 Model presets for YOLO, MobileNet, EfficientNet, ResNet, ViT, CLIP, SAM, DINO, DETR
• 🎯 Native quantization: Float→Int8/Uint8/Int16 with per-tensor and per-channel support (TFLite compatible)
• 📦 Batch processing with concurrency control
• 🧮 Tensor operations: channel extraction, patch extraction, permutation, batch concatenation

Image Analysis & Augmentation

• 📈 Image analysis: statistics, metadata, validation, blur detection
• 🔄 Augmentation: rotation, flip, brightness, contrast, saturation, blur
• 🔙 Tensor to image conversion

Computer Vision Utilities

• 🏷️ Built-in label databases: COCO, ImageNet, VOC, CIFAR, Places365, ADE20K
• 📹 Camera frame processing: YUV/NV12/BGRA→tensor conversion
• 📦 Bounding box utilities: format conversion, scaling, clipping, IoU, NMS
• 🖼️ Letterbox padding with reverse coordinate transform
• 🎨 Drawing/visualization: boxes, keypoints, masks, heatmaps

v1.2.1

Add cutout/random erasing augmentation for robustness training

• Add CutoutOptions and CutoutResult TypeScript types
• Support multiple cutouts with configurable size and aspect ratio ranges
• Three fill modes: constant color, noise, random color per cutout
• Probability parameter for probabilistic application
• Optional seed for reproducible random erasing
• iOS implementation with CoreGraphics noise generation
• Android implementation with Bitmap pixel manipulation
• Returns detailed region info (x, y, width, height, fill)
• Add comprehensive documentation in README
• Add example app integration with Cutout button
• Add unit tests for all cutout scenarios
• Update jest mocks for cutout
• Bump version to 1.2.1