In this episode of Mobile Development with Fexingo, Lucas and Luna explore how on-device AI is transforming handwriting recognition on smartphones, moving beyond basic OCR to real-time, stroke-level analysis. They focus on a specific case: the MyScript Nebo app for iPad and iPhone, which uses neural networks to convert handwritten notes into digital text instantly, without sending data to the cloud. Lucas explains how the technology works—using a recurrent neural network trained on millions of handwriting samples to predict characters as you write—and why latency under 50 milliseconds is critical for a natural feel. Luna raises the challenge of handling diverse handwriting styles and the privacy benefits of on-device processing. The hosts also touch on Apple's PencilKit integration and how developers can leverage Core ML to build similar features. The episode closes with a look at future applications, such as real-time equation solving and sketch recognition.

#OnDeviceAI #HandwritingRecognition #MobileDevelopment #iOS #Android #ApplePencil #CoreML #MyScript #Nebo #NeuralNetworks #OCR #RealTimeAI #Privacy #AppDevelopment #Technology #MachineLearning #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode, Lucas and Luna explore how mobile apps use on-device AI to track trampoline jumps and flips in real time. They break down the specific sensors involved—accelerometer, gyroscope, and magnetometer—and how machine learning models classify movement patterns like continuous vertical oscillation versus a controlled flip. Lucas shares data from a study showing that jump-counting accuracy drops from 97% to 82% when a phone is in a pocket versus a wristband. They also discuss the privacy gains of processing all data locally, and how this tech could extend to other sports like cheerleading or parkour. Luna brings up the unexpected challenge of distinguishing a real bounce from a shaky car ride. The episode includes a brief, natural mention of listener support through Buy Me a Coffee.

#OnDeviceAI #MobileApps #TrampolineTracking #RealTimeAI #MachineLearning #SensorFusion #Accelerometer #Gyroscope #JumpCounting #SportsTech #FitnessTracking #Privacy #EdgeAI #SmartphoneSensors #Technology #FexingoBusiness #BusinessPodcast #MobileDevelopment

Keep every episode free: buymeacoffee.com/fexingo

Smartphones today can spot a sick tomato plant before your eyes do. In this episode, Lucas and Luna explore how on-device AI is being deployed for real-time plant disease detection — from cassava in East Africa to vineyards in California. They break down the technical challenge: training lightweight convolutional neural networks on leaf image datasets, achieving 95% accuracy in under 200 milliseconds without a cloud round trip. The hosts discuss a specific case: PlantVillage's Nuru app, which runs entirely offline on a $50 smartphone and has been deployed by over 100,000 smallholder farmers in Kenya and Tanzania. They also examine the hardware constraints — Qualcomm's Snapdragon neural processing units, Apple's Neural Engine, and Google's Tensor Processing Unit in Pixel phones — and why latency matters more for a farmer in a field than for a tech demo. A concrete look at how edge AI is changing agriculture, one leaf at a time.

#PlantDiseaseDetection #OnDeviceAI #MobileApps #Agriculture #Agtech #ComputerVision #DeepLearning #EdgeAI #Farming #Smartphone #PlantVillage #Nuru #Cassava #Snapdragon #NeuralEngine #TPU #Technology #FexingoBusiness

Keep every episode free: buymeacoffee.com/fexingo

In this episode of Mobile Development with Fexingo, Lucas and Luna dive into the fascinating world of on-device AI for lip-reading. They explore how modern smartphones can transcribe speech from silent video by analyzing mouth movements, using lightweight neural networks like LipNet. The hosts discuss the technical challenges of training on datasets like GRID and LRW, the privacy advantages of processing everything on-device, and real-world use cases from accessibility for the hearing impaired to silent command input. They also touch on the trade-offs between accuracy and speed, and how this technology is evolving with Apple's Neural Engine and TensorFlow Lite. Tune in for a deep dive into a cutting-edge mobile AI application that's quietly reshaping how we interact with our devices.

#OnDeviceAI #LipReading #MobileAI #Accessibility #LipNet #NeuralNetworks #TensorFlowLite #AppleNeuralEngine #DeepLearning #ComputerVision #SpeechRecognition #MobileDevelopment #iOS #Android #Privacy #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

Episode 95 of Mobile Development with Fexingo dives into acoustic scene classification — on-device AI that identifies whether you're in a park, a subway, or a busy restaurant by analyzing ambient sound. Lucas and Luna break down how Google's Sound Events detection model (part of ML Kit) processes audio in milliseconds without sending data to the cloud. They explore a real-world case: a hearing aid app called 'Soundscape' that adjusts filtering based on detected environment, improving user experience while preserving privacy. The hosts discuss the technical challenges — training data bias, model size constraints on mobile, and battery impact — and why this capability is becoming a standard feature in health and accessibility apps. If you've ever wondered how your phone knows you're in a coffee shop before you pull it out, this episode explains the machine learning behind it.

#AcousticSceneClassification #OnDeviceAI #MobileAI #GoogleMLKit #SoundEvents #Soundscape #HearingAidApp #AccessibilityTech #HealthTech #AmbientSound #AudioAI #Privacy #MachineLearning #MobileDevelopment #iOSDev #AndroidDev #FexingoBusiness #Technology

Keep every episode free: buymeacoffee.com/fexingo

Lucas and Luna explore how on-device AI is enabling real-time lip reading on smartphones. They discuss the core technology—visual speech recognition using lightweight neural networks—and highlight a real-world case: Google's Project Euphonia spin-off that helps people with speech impairments communicate via silent mouthing. The episode covers the technical challenges like lighting and angle robustness, the trade-offs between accuracy and latency, and why running inference entirely on-device is critical for privacy. Lucas breaks down the model architecture (CNN + Transformer) and how it's optimized for mobile Neural Processing Units. Luna asks the listener's questions about training data requirements and real-world accuracy. A concrete episode for anyone interested in the cutting edge of mobile AI.

#LipReading #OnDeviceAI #VisualSpeechRecognition #MobileAI #GoogleProjectEuphonia #NeuralProcessingUnit #CNN #Transformer #PrivacyPreservingAI #RealTimeAI #AccessibilityTech #SpeechImpairment #SilentSpeech #MachineLearning #DeepLearning #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

Episode 93 of Mobile Development with Fexingo explores how on-device AI is turning smartphones into driver drowsiness detectors. Lucas and Luna break down the specific sensors and models involved: the forward-facing camera, infrared flood illuminator, and a lightweight convolutional neural network that tracks eye aspect ratio and head pose in real time. They discuss the technical challenge of running inference under 20 milliseconds while preserving battery life, and how Apple's Core ML and Google's ML Kit enable this. The episode also touches on the regulatory push from the EU's General Safety Regulation and how automakers like Volvo are integrating phone-based alerts with in-car systems. A concrete look at the intersection of mobile development, computer vision, and road safety.

#DriverDrowsiness #OnDeviceAI #ComputerVision #CoreML #MLKit #MobileDevelopment #iOSDev #AndroidDev #SafetyTech #AutomotiveAI #RoadSafety #EUGeneralSafetyRegulation #Volvo #SmartphoneSensors #NeuralNetworks #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode of Mobile Development with Fexingo, Lucas and Luna dive into on-device depth estimation — how modern smartphones use AI to understand 3D space without cloud servers. They focus on Apple's LiDAR scanner and Google's Depth API, explaining how these technologies power features like portrait mode, AR measurement apps, and real-time object occlusion. Lucas breaks down the difference between structured light, time-of-flight, and monocular depth estimation, using the example of the IKEA Place app to show how accurate depth data lets you virtually place furniture in your room. They also discuss the challenges of computational cost and battery life, and how developers can balance accuracy with performance using Apple's Metal Performance Shaders and Google's MediaPipe. A concrete takeaway: the latest flagship phones can estimate depth at 30 frames per second using less than 5% of the CPU, thanks to dedicated neural processing units.

#MobileApps #OnDeviceAI #DepthEstimation #ComputerVision #AppleLiDAR #GoogleDepthAPI #AR #AugmentedReality #MachineLearning #NeuralNetworks #iOS #Android #Technology #FexingoBusiness #BusinessPodcast #MobileDevelopment #AppDevelopment #RealTimeAI

Keep every episode free: buymeacoffee.com/fexingo

Episode 91 of Mobile Development with Fexingo dives into on-device AI for real-time skin tone matching in mobile apps. Lucas and Luna explore how companies like L'Oréal and Google have integrated inclusive skin tone sensors into smartphone cameras, enabling accurate foundation shade matching and personalized beauty recommendations without sending data to the cloud. They discuss the technical challenges: training models on diverse skin tones using the Monk Skin Tone scale, optimizing for low latency on devices, and avoiding bias in training data. The episode also touches on privacy benefits of on-device processing and how this technology is expanding into health and fitness wearables. A concrete look at how AI is making mobile apps more inclusive and private.

#SkinToneMatching #OnDeviceAI #MobileApps #BeautyTech #InclusiveAI #Loreal #Google #MonkSkinTone #ComputerVision #EdgeAI #PrivacyByDesign #FacialRecognition #ARMakeup #SmartphonePhotography #DeepLearning #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode, Lucas and Luna dive into the world of computational photography on smartphones, specifically focusing on how on-device AI is now capable of removing lens flares in real time. They explore the technical challenges of distinguishing flares from legitimate light sources, the hardware accelerators like Apple's Neural Engine and Google's Tensor Processing Unit that make this possible, and the practical implications for casual photographers and social media content creators. The discussion also touches on how this technology compares to traditional post-processing tools and what it means for the future of mobile photography. A concrete example is given of a recent update to the camera app on a popular flagship phone that reduced flare removal processing time from 2 seconds to under 200 milliseconds.

#OnDeviceAI #ComputationalPhotography #LensFlareRemoval #MobilePhotography #AppleNeuralEngine #GoogleTensor #AIProcessing #CameraApp #RealTimeAI #SmartphonePhotography #ImageProcessing #DeepLearning #NeuralNetworks #TechInnovation #MobileTech #FexingoBusiness #BusinessPodcast #Technology

Keep every episode free: buymeacoffee.com/fexingo