TensorFlow Lite
PulseAugur coverage of TensorFlow Lite — every cluster mentioning TensorFlow Lite across labs, papers, and developer communities, ranked by signal.
4 day(s) with sentiment data
-
New loss function corrects green skin-tone casts in lightweight AI photo editors
Researchers have identified a subtle but persistent green color cast introduced by ultra-lightweight photo retouching networks, particularly affecting human skin tones. This issue arises even when models are trained usi…
-
New SDK enables streaming speech enhancement on diverse platforms
Researchers have introduced GigaEdge Spench, an open-source SDK designed for streaming speech enhancement across various platforms including servers, mobile devices, and web browsers. The SDK allows applications to cons…
-
Low-cost AgriDefend device uses AI to detect crop stress before symptoms appear
Researchers have developed AgriDefend, a low-cost offline edge device designed to detect crop stress before visible symptoms appear. This device fuses data from canopy imagery analyzed by a MobileNetV2 neural network an…
-
Samsung uses TensorFlow Lite for image-based tracking
Samsung is reportedly using a TensorFlow Lite model to track individuals via images captured at a 960x540 resolution. This application of AI technology raises privacy concerns regarding personal tracking through visual data.
-
New SIDE method detects sensor impersonation in IoT devices
Researchers have developed a novel method called SIDE for detecting sensor impersonation in Internet of Things (IoT) devices. This approach treats detection as a sequence prediction problem, utilizing a lightweight mode…
-
Deep Microcompression enables CNNs on microcontrollers
A new paper introduces Deep Microcompression (DMC), a method designed to optimize deep learning models for microcontrollers. DMC combines structured pruning, quantization-aware training, and bit-packing to significantly…
-
AI system adapts dehazing for clearer vision in smoke
Researchers have developed an input-adaptive gating system for a dehazing front-end designed for on-device perception in smoke-obscured environments. This system, tested on a Raspberry Pi 4 Model B using TensorFlow Lite…
-
Flutter developer seeks help with TFLite model errors after image resizing
A user is encountering significant errors when integrating a TensorFlow Lite (TFLite) model, originally trained using MobileNetv3, into a Flutter application. The issue arises after the camera stream captures images, wh…
-
AI system developed for on-device malaria detection with clinical focus
Researchers have developed an on-device system for detecting malaria using microscopy images, addressing critical clinical requirements often overlooked in machine learning literature. The system incorporates stopping c…
-
Deep learning models enhance H-IoT cybersecurity with lightweight detection
Researchers have developed new deep learning models, Temporal Convolutional Network (TCN) and Residual TCN (Res-TCN), to enhance cybersecurity for healthcare Internet of Things (H-IoT) systems. These lightweight models …
-
LLMs drive neural architecture search with new methods for code and mobile deployment
Two new research papers explore the use of Large Language Models (LLMs) in Neural Architecture Search (NAS). The first paper, 'GraphIR', introduces an intermediate representation to bridge the gap between executable neu…
-
New framework FBLayout boosts mobile LLM fine-tuning efficiency
Researchers have developed FBLayout, a new framework designed to optimize memory layout for efficient fine-tuning of large language models on mobile GPUs. This approach addresses the memory constraints and layout transf…
-
Google launches LiteRT.js for in-browser AI model inference
Google has introduced LiteRT.js, a JavaScript binding for its on-device inference library, formerly known as TensorFlow Lite. This new tool allows .tflite models to run directly within web browsers, leveraging WebGPU, W…
-
Run Neural Networks on GPU via Vulkan: Libraries, Compilers, or Custom Engines
This article outlines three methods for running trained neural networks on a GPU using the Vulkan API. It suggests integrating existing libraries like TensorFlow Lite or ONNX Runtime, compiling models via ML compilers s…
-
PrintGuard 2.0 launches with 5MB TFLite model for browser and CPython
PrintGuard 2.0 is an updated system for detecting failures in 3D printing, utilizing a ShuffleNetV2 encoder and a prototypical network for few-shot fault detection. The new version features a significantly smaller Tenso…
-
AI mobile guide for Grand Egyptian Museum developed
Researchers have developed TimeLens, an AI-powered mobile guide for the Grand Egyptian Museum. This system can recognize artifacts in real-time and answer visitor questions in English or Arabic. The project involved cre…
-
Tiny collaborative inference boosts object detection on edge devices
Researchers have developed a method for improving object detection on small edge devices, particularly in scenarios with occlusion. Their approach combines a lightweight neural network architecture with TensorFlow Lite …
-
DCGAN runs on RISC-V microcontroller with 512KB SRAM
A project successfully implemented a 12.6 million parameter DCGAN model for generating 64x64 cat faces on a dual-core RISC-V microcontroller with only 512KB of SRAM. The inference engine, written entirely in C, achieved…
-
Deep learning ensemble boosts plant disease classification accuracy
Researchers have developed AgriMind, an ensemble deep learning framework designed to automate plant disease classification. This system combines three models—ResNet50, EfficientNet-B0, and DenseNet121—trained on over 20…
-
AI model grades knee osteoarthritis severity on limited devices
Researchers have developed a novel approach for grading knee osteoarthritis severity using a combination of deep learning and a large language model. The system utilizes a ResNet-18 convolutional neural network, optimiz…