In episode 68, Lucas and Luna explore how mobile apps are using on-device AI to colorize black-and-white photos in real time. They dive into the technology behind this feature—from convolutional neural networks trained on millions of colorized images to the challenge of maintaining historical accuracy. Lucas shares how Apple and Google have integrated colorization into their native photo editing tools on iPhone and Android, while Luna discusses third-party apps like Colorize AI and Palette that use neural networks running entirely on the device, with no cloud upload required. The episode covers the key trade-off between speed and fidelity, how mid-range Android phones handle the processing load, and why this capability matters for preserving family history and creative editing. They also touch on how developers optimize models for edge inference using frameworks like TensorFlow Lite and Core ML. A concrete example: the MyHeritage app colorized 30 million old photos in its first year, with 95% of processing done on-device by 2026. By the end, listeners will understand the pipeline from grayscale input to vibrant output, and how the industry is pushing toward sub-second colorization for live camera preview.

#AI #OnDeviceAI #MobileApps #PhotoColorization #MachineLearning #NeuralNetworks #Apple #Google #TensorFlowLite #CoreML #iPhone #Android #RealTimeColorization #HistoryPreservation #DeepLearning #MobilePhotography #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode of Mobile Development with Fexingo, Lucas and Luna explore how modern mobile apps leverage on-device AI to transform document scanning from a simple photo capture into a powerful tool for text extraction, formatting, and search. They dive into Apple's Vision framework, Google's ML Kit, and real-world apps like Adobe Scan and Microsoft Lens, discussing how these technologies enable features like auto-cropping, perspective correction, optical character recognition, and even handwriting recognition — all without sending data to the cloud. The hosts break down the trade-offs between on-device and cloud-based processing, including privacy, speed, and accuracy, and touch on the latest advancements in 2026 that make on-device scanning faster and more reliable than ever. Along the way, they share practical tips for developers looking to integrate these capabilities into their own apps. A must-listen for any mobile developer interested in computer vision and privacy-preserving AI.

#MobileDevelopment #OnDeviceAI #DocumentScanning #ComputerVision #OCR #AppleVision #MLKit #AdobeScan #MicrosoftLens #TextRecognition #HandwritingRecognition #Privacy #MachineLearning #iOS #Android #FexingoBusiness #BusinessPodcast #Technology

Keep every episode free: buymeacoffee.com/fexingo

In this episode of Mobile Development with Fexingo, Lucas and Luna explore how on-device AI powers real-time food recognition in mobile apps. They dive into how Calorie Mama and Snapchat's Dish Scanner use on-device machine learning to identify dishes from a photo, estimate nutritional content, and even suggest recipes. Lucas breaks down the technical stack: lightweight convolutional neural networks fine-tuned for food categorization, running entirely on the device via Apple's Core ML and Google's ML Kit. Luna shares data on accuracy rates—around 85% for common dishes—and how models are trained on the Food-101 dataset with 101,000 images. They discuss the shift from cloud-based to on-device inference, highlighting speed and privacy benefits, and touch on challenges like portion size estimation and cross-cultural cuisines. Tune in for a concrete look at how smartphone cameras become dietary assistants without sending your lunch to a server.

#MobileDevelopment #OnDeviceAI #FoodRecognition #ComputerVision #CoreML #MLKit #CalorieMama #Snapchat #Food101 #MachineLearning #DeepLearning #iOSDevelopment #AndroidDevelopment #Privacy #RealTimeAI #FexingoBusiness #BusinessPodcast #Technology

Keep every episode free: buymeacoffee.com/fexingo

Episode 65 of Mobile Development with Fexingo dives into on-device AI for real-time 3D body mesh. Lucas and Luna break down how apps like Snapchat and FitOn use neural networks to map thousands of body points without sending data to the cloud. They discuss the tech stack — MediaPipe, ARKit, and TensorFlow Lite — and the engineering trade-offs between accuracy and battery life. Specific focus: how the latest Qualcomm Snapdragon chip enables sub-10-millisecond mesh inference on mid-range phones. Also covered: privacy implications, the jump from 2D skeleton tracking to volumetric mesh, and why this matters for fitness, fashion, and AR try-ons. No ads — listener-supported at buy me a coffee dot com slash fexingo.

#3DBodyMesh #OnDeviceAI #MobileApps #AR #Snapchat #MediaPipe #ARKit #TensorFlowLite #Qualcomm #Snapdragon #FitnessApp #FashionTech #ComputerVision #Privacy #EdgeAI #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode of Mobile Development with Fexingo, Lucas and Luna dive into how modern smartphones leverage on-device AI to optimize camera exposure in real time—no cloud processing required. They break down the mechanics behind Apple's Smart HDR and Google's Real Tone, explaining how neural networks analyze scene luminance, subject motion, and color temperature in milliseconds. The discussion covers the shift from traditional histogram-based metering to AI-driven patch-wise exposure, the role of neural processing units (NPUs) like the A18 Bionic and Tensor G4, and why this matters for app developers building camera features. Specific examples include how the iPhone 16 Pro handles high-contrast backlit portraits and how Pixel 10's AI avoids blowing out highlights in concert photography. The hosts also touch on the challenges of training models on diverse skin tones and lighting conditions. A must-listen for any mobile developer interested in computational photography.

#OnDeviceAI #CameraExposure #MobilePhotography #SmartHDR #RealTone #Apple #Google #NPU #A18Bionic #TensorG4 #IPhone16Pro #Pixel10 #ComputationalPhotography #MachineLearning #iOSDevelopment #AndroidDevelopment #Tech #FexingoBusiness

Keep every episode free: buymeacoffee.com/fexingo

In this episode, Lucas and Luna dive into the on-device AI behind real-time background blur on smartphones. They explore how Apple, Google, and Samsung implement portrait mode for video calls and live streaming, using neural networks and depth sensors. The discussion covers the evolution from dual-camera setups to single-camera solutions with machine learning, the role of the Neural Engine and Tensor Processing Unit, and how apps like Zoom and Instagram leverage these features. Specific numbers include the 15 billion operations per second required for real-time blur at 30 frames per second on a flagship chip. Lucas explains how the model is trained on synthetic data and fine-tuned for edge devices, while Luna questions the trade-offs in accuracy and power consumption. The episode also touches on the future of computational photography and the impact on privacy, since processing happens entirely on-device. No prior episode has covered this specific angle of on-device AI for real-time background blur in mobile apps.

#OnDeviceAI #BackgroundBlur #PortraitMode #ComputationalPhotography #MobileApps #Apple #Google #Samsung #NeuralEngine #TensorProcessingUnit #Zoom #Instagram #DepthMapping #MachineLearning #RealTimeVideo #SmartphoneCamera #Technology #FexingoBusiness

Keep every episode free: buymeacoffee.com/fexingo

Episode 62 of Mobile Development with Fexingo dives into the nuts and bolts of on-device HDR photo stitching. Lucas and Luna break down how modern smartphones capture three separate exposures in under 100 milliseconds, align them pixel-by-pixel using neural networks, and merge them without sending a single image to the cloud. They walk through the actual computational pipeline — from raw sensor readout to tone mapping — and explain why Google's 2026 Pixel approach differs from Apple's. The episode also touches on the trade-off between dynamic range and natural-looking highlights, and why the best HDR often looks like it wasn't HDR at all.

#HDR #OnDeviceAI #MobilePhotography #ComputationalPhotography #GooglePixel #AppleiPhone #ToneMapping #NeuralNetworks #SmartphoneCamera #ImageSignalProcessor #ZeroShutterLag #BracketedExposure #RealTimeProcessing #Technology #MobileDevelopment #FexingoBusiness #BusinessPodcast #ArtificialIntelligence

Keep every episode free: buymeacoffee.com/fexingo

Depth mapping is the invisible layer that powers portrait mode, AR measuring tools, and gesture control on your phone. In this episode, Lucas and Luna unpack how LiDAR and neural networks turn a 2D camera feed into a 3D understanding of the world. They walk through Apple's integration of a dot projector and flood illuminator on iPhones starting in 2017, and how the latest 2026 Android flagships from Honor and Xiaomi use dual-spectrum sensors to achieve sub-millimeter depth accuracy. The conversation covers practical use cases: from real-time furniture placement in IKEA's app to assistive navigation for visually impaired users via Microsoft's Seeing AI. They also touch on the privacy implications of offloading depth data to the neural engine rather than the cloud. A focused look at one sensor technology that quietly reshapes mobile interactions.

#MobileDepthMapping #OnDeviceAI #LiDAR #Apple #Android #Honor #Xiaomi #IKEA #Microsoft #SeeingAI #PortraitMode #AR #NeuralEngine #Privacy #Technology #FexingoBusiness #BusinessPodcast #MobileDevelopment

Keep every episode free: buymeacoffee.com/fexingo

In episode 60 of Mobile Development with Fexingo, Lucas and Luna explore how on-device AI is enabling real-time emotion recognition in mobile apps. They dive into the technical backbone—facial action coding systems running locally on smartphones—and discuss practical use cases from mental health apps to adaptive gaming. The hosts examine how companies like Affectiva and Apple are approaching this sensitive technology, the privacy advantages of on-device processing, and the ethical questions around emotional data. They also touch on the accuracy challenges and the future of emotion-aware interfaces, all without relying on cloud connectivity. A focused, insightful conversation for mobile developers and tech enthusiasts.

#EmotionRecognition #OnDeviceAI #MobileApps #FacialActionCoding #Affectiva #Apple #MentalHealthTech #AdaptiveGaming #Privacy #MachineLearning #CoreML #ARKit #RealTimeAI #EthicalAI #MobileDevelopment #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

Lucas and Luna dive into how modern mobile apps use on-device AI for real-time barcode and QR code scanning, beyond simple price checks. They explore the underlying machine learning models that enable instant recognition of damaged or partially obscured codes, and how apps like those for inventory management and contactless payments leverage dedicated neural processing units (NPUs) on recent iPhones and Android devices. The conversation covers the shift from traditional image-processing algorithms to deep learning models that handle low-light conditions, curved surfaces, and dynamic environments. Specific examples include retail inventory apps that use real-time scanning to count stock without manual entry, and how the Google Lens framework processes QR codes for augmented reality experiences. They also discuss privacy implications: on-device processing means codes are never sent to the cloud, which is critical for payment or personal data. By mid-2026, Apple's Core ML and Google's ML Kit offer pre-trained models that developers can integrate in minutes, making high-accuracy scanning accessible to any app. Listeners will learn one concrete improvement: modern on-device models can decode a QR code in under 50 milliseconds, even when 30% of the code is damaged.

#OnDeviceAI #MobileApps #BarcodeScanning #QRCode #MachineLearning #iOS #Android #CoreML #MLKit #NPU #RealTime #Privacy #InventoryManagement #ContactlessPayments #AugmentedReality #GoogleLens #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo