A developer released sentinel-imxd, a lightweight log-anomaly detection daemon for the NXP i.MX 8M Plus that runs a TensorFlow Lite autoencoder on the board's Vivante NPU via the VX external delegate.…
Google's LiteRT, the renamed TensorFlow Lite, has evolved with new package names and APIs, including the CompiledModel interface and litert-torch for PyTorch conversion. The TensorFlow Lite packages a…
A developer detailed how to build a real-time physical therapy form-correction app using React Native and MediaPipe pose estimation. The tutorial covers on-device landmark detection, joint angle calcu…
A developer's team at an unnamed company built a predictive autoscaling system for GPU workloads on Kubernetes after a production outage caused by reactive scaling. The system uses a Bi-LSTM model emb…
A developer outlines five common causes of frame drops in real-time camera filters, including format conversion, frame backpressure, inference execution location, synchronous readback, and thermal loa…
Google's LiteRT.js, a browser runtime for on-device AI inference using WebAssembly and WebGPU, enables developers to run standard .tflite models directly in the browser, as demonstrated in a tutorial …
A developer has open-sourced offline_sos_system, a pure Dart crash detection engine for Flutter that runs entirely on-device using TensorFlow Lite, eliminating the need for cloud APIs or network conne…
Google has introduced an AI model that estimates body fat from a single selfie, with a mean absolute error of about 3.8% across BMI 18–35, using a MobileNetV3-small backbone with 1.2 million parameter…
A developer detailed a method for implementing object detection on Android devices to enable autonomous robots to recognize objects locally, reducing reliance on network connectivity. The approach use…
A developer detailed how to wire MediaPipe's CalculatorGraph API to custom TensorFlow Lite models on Android via a Kotlin/JNI bridge, achieving sub-16ms streaming inference on mid-range Snapdragon har…
A developer detailed the architecture for building an on-device LLM chatbot using Kotlin and TensorFlow Lite, focusing on mobile integration, model execution, prompt handling, and performance. The tut…
A new arXiv preprint (2608.00078v1) presents a mobile deployment pipeline that uses QLoRA fine-tuning of an LLM to generate convolutional neural networks, with INT8 TensorFlow Lite export and on-devic…
STMicroelectronics (ST) announced the STM32N6, its most powerful microcontroller and the first to include a custom Neural-ART Accelerator NPU capable of 600 GOPS, enabling machine learning application…
A developer compares PyTorch and TensorFlow for computer vision projects, noting that PyTorch dominates research with dynamic computation graphs and recent performance improvements via torch.compile()…
A developer details how to build high-performance edge AI pipelines on Android using CameraX and TFLite, addressing common performance pitfalls like the 'Edge AI Wall' and advocating for hardware acce…
Google's AICore transforms AI models from app-bundled assets into system services on Android, enabling heterogeneous parallelism across NPU, GPU, and DSP. The approach addresses the memory wall via ze…
Google's MediaPipe Tasks and AICore are revolutionizing on-device machine learning for Android developers by abstracting low-level tensor manipulation into declarative pipeline orchestration. MediaPip…
Google released LiteRT-LM, a new runtime framework that accelerates on-device inference for its Gemma 4 large language models by up to 2.2x using multi-token prediction. The framework, built on the Li…
Researchers have developed a collaborative inference method for occlusion-robust object detection on edge devices with less than 1 MB SRAM, combining an MCUNet backbone, YOLOv2 detection head, and Ten…
A Swedish farm developer has ordered a single PIR motion sensor as the first component of FenceGuard, an open-source early warning system for livestock predators. The project deliberately starts with …