qualcomm/ai-hub-models

Qualcomm® AI Hub Models is our collection of state-of-the-art machine learning models optimized for performance (latency, memory etc.) and ready to deploy on Qualcomm® devices.

What it solves

It provides a curated collection of state-of-the-art machine learning models specifically optimized for deployment on Qualcomm devices, reducing the friction of compiling, quantizing, and profiling models for on-device hardware.

How it works

Users can browse and download models via a CLI or Python API. By integrating with the Qualcomm AI Hub Workbench, the project allows users to compile models for specific target runtimes (like TFLite or ONNX) and hardware targets (CPU, GPU, NPU), perform quantization, and profile performance on real cloud-hosted devices before downloading the final deployable asset.

Who it’s for

Developers building AI applications for Android, Linux, or Windows devices powered by Snapdragon chipsets and other Qualcomm platforms.

Highlights

Broad Hardware Support: Optimized for a wide range of Snapdragon Mobile, Compute, Automotive, and IoT platforms.
Comprehensive Model Zoo: Includes a vast array of models for image classification, image editing, super-resolution, and semantic segmentation.
End-to-End Demos: Provides CLI demos that handle preprocessing, inference, and postprocessing for most models.
Multi-Runtime Support: Compatible with Qualcomm AI Engine Direct, LiteRT (TensorFlow Lite), and ONNX.