qualcomm-npu

workbench for learning and practicing on-device AI technologies under real-world scenarios on Android smartphones with online TV. powered by llama.cpp + whisper.cpp + FFmpeg + opencv-mobile

the original reference implementation of a specified llama.cpp backend for Qualcomm Hexagon NPU on Android phone, history of ggml-hexagon: https://github.com/zhouwg/ggml-hexagon/discussions/18.

On-device CLIP inference on Android with ExecuTorch + Qualcomm QNN backend. Zero-shot image classification and vision-language understanding powered by hardware-accelerated AI on Snapdragon devices.

Catalog of projects that build upon Qualcomm's platforms

Lightweight UAV-SAR perception system for aerial human detection, monocular GPS geolocation, and automated rescue alerting using standard UAV telemetry and edge AI.

QNN LLM binding for React Native

Unleash your NPU on Linux

Qualcomm lib Genie binding for Node.js

Description: On-device Android AI assistant for face recognition, object detection, speech I/O, and memory-aware assistance using ExecuTorch, TFLite, Whisper, and Piper.

Complete guide and scripts for setup, execution and testing the AI models (ONNX) performance on NPU architectures (Qualcomm QNN & Intel OpenVINO) on Windows.

Lightweight underwater pipeline perception using YOLOv8n segmentation, PCA-based geometry extraction, and embedded AI deployment.

Parents already film their kids' cute walks. ToddleAI flags gait & posture concerns from those videos using on-device AI, early detection and action.

Showcase repository for MobiDeflare: a deep learning-based Android application for on-device, real-time lens flare removal.

Repository containing source code to reproduce results from article