Open Source Machine Learning Software

Gallery is a curated collection of on-device machine learning examples, demo apps, and model artifacts designed to help developers experiment with and deploy ML at the edge. The project bundles runnable samples that show how to run TensorFlow Lite/Edge TPU models (and similar lightweight runtimes) on mobile and embedded platforms, demonstrating common tasks like image classification, object detection, audio recognition, and pose estimation. Each sample is intended to be both a learning aid and a practical starting point: code is organized to show model loading, pre/post-processing, performance measurement, and common optimization knobs (quantization, NNAPI/Delegate usage, and hardware accelerators). The repo also collects small, well-documented models and conversion scripts so developers can reproduce a pipeline from a full-size model down to a device-friendly artifact.

OpenCV (Open Source Computer Vision Library) is a comprehensive open-source library for computer vision, machine learning, and image processing. It enables developers to build real-time vision applications ranging from facial recognition to object tracking. OpenCV supports a wide range of programming languages including C++, Python, and Java, and is optimized for both CPU and GPU operations.

LiteRT-LM is Google’s open-source inference framework for deploying large language models on edge devices. It is built for production-oriented local LLM execution across Android, iOS, desktop, web, embedded, and IoT environments. The framework focuses on performance, hardware acceleration, and efficient model serving close to the user instead of relying only on remote cloud inference. It supports CPU execution across major platforms and adds GPU or NPU acceleration where available. LiteRT-LM is especially relevant for developers building private, low-latency AI features on phones, laptops, Raspberry Pi-style devices, and other edge hardware. Its goal is to make modern language models usable in local applications with a consistent deployment stack.

Adaptive Intelligence is the implementation of neural science, forensic psychology , behavioral science with machine-learning and artificial intelligence to provide advanced automated software platforms with the ability to adjust and thrive in dynamic environments by combining cognitive flexibility, emotional regulation, resilience, and practical problem-solving skills.

nunn is a collection of ML algorithms and related examples written in modern C++17.

There are many tools available in market which will provide spell correction or grammer correction while making documents, but very few tools are available which are providing sentence completion according to previously entered text. But this all are providing sentence complition suggestion for sentences which are oftenly or regularly used by all people in same manner. But in reality style of writing changes person to person. While our aim is to provide a sentence suggestion tool which will give suggestion to complete the sentence according previously enterd data by the user. Output or suggestion for same sentence or word will change person to person according to previously entered data by that person. So, it will be very easy to type any document, sms, mail, chatting etc.

Some of the usual problems for Learning vector quantization (LVQ) based methods are that one cannot optimally guess about the number of prototypes required for initialization for multimodal data structures i.e.these algorithms are very sensitive to initialization of prototypes and one has to pre define the optimal number of prototypes before running the algorithm. If a prototype, for some reasons, is ‘outside’ the cluster which it should represent and if there are points of a different categories in between, then the other points act as a barrier and the prototype will not find its optimum position during training. Since the model complexity is not known in many cases, we avoid this problem by introducing a "Dynamic" version of LVQ. Dynamic-GRLVQ (DGRLVQ), which adapts the model complexity to the given problem during training by adding or removing prototypes dynamically/realtime one by one for each category until satisfactory classification results are achieved.

KotlinDL is a high-level Deep Learning API written in Kotlin and inspired by Keras. Under the hood, it uses TensorFlow Java API and ONNX Runtime API for Java. KotlinDL offers simple APIs for training deep learning models from scratch, importing existing Keras and ONNX models for inference, and leveraging transfer learning for tailoring existing pre-trained models to your tasks. This project aims to make Deep Learning easier for JVM and Android developers and simplify deploying deep learning models in production environments.

LOSS is a high-speed, "split-stack" financial dashboard designed to strip away market noise. By combining mathematically rigorous technical indicators with VADER-powered AI sentiment analysis, LOSS delivers objective, real-time Buy, Sell, or Hold verdicts in under two seconds. Wrapped in a distraction-free, "Brutalist" user interface, this tool is built for traders who rely on data, not hype. Modern financial websites are bloated with ads, conflicting opinions, and unnecessary jargon. LOSS solves this by automating the math and reading the news for you. Instead of a monolithic architecture, LOSS uses a Split-Stack approach: A Vercel Relay handles raw data fetching, bypassing CORS and rate limits. A Render Python Backend acts as the "Brain," executing heavy machine learning and NLP tasks. Try the live application in action here: https://loss-by-vignesh.vercel.app before downloading the source code Visit For More Details : https://github.com/Vignesh-72/Loss

libcrn is document image processing library written in C++11 for Linux, Windows, Mac OsX and Google Android. It is a toolbox that allows to create easily software such as OCRs and layout analysis tools.