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 Qu
What is 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.
Overview of ai-hub-models
What it is
h3>Qualcomm® AI Hub Models is a collection of state-of-the-art machine learning models optimized for performance (latency, memory, etc.) and ready to deploy on Qualcomm® devices. The collection includes models for generative AI, image generation, text generation, and other categories. Each model is accompanied by on-device performance data, device-native sample apps, and a command-line interface (CLI) for browsing, downloading, and running models. The project is open source under the BSD-3-Clause license. h3>
Key features
h3>
- State-of-the-art ML models optimized for deployment on Qualcomm® devices. h3>
- On-device performance data, including latency and memory metrics. h3>
- Device-native sample applications for running models on physical devices. h3>
- A CLI (
qai-hub-models) for browsing the catalog, fetching deployable assets, and running end-to-end demos. h3> - Support for multiple on-device runtimes: Qualcomm AI Engine Direct, LiteRT (TensorFlow Lite), and ONNX. h3>
- Hardware targets include CPU (FP32, INT16, INT8), GPU (FP32, FP16), and NPU (FP16, INT16, INT8), across a range of Snapdragon and Qualcomm chipsets. h3>
- Ability to compile, export, quantize, and profile models on cloud-hosted Qualcomm® devices via AI Hub Workbench. h3>
Use cases
h3>
- Deploying generative AI models (text and image generation) on Qualcomm® devices for edge inference. h3>
- Running computer vision models (e.g., object detection, classification) on smartphones, automotive platforms, and IoT devices. h3>
- Prototyping and evaluating model performance on real hardware before production deployment. h3>
- Developing native or Python applications that incorporate pre/post-processing and accelerated inference. h3>
Getting started
h3> h3>See the repository README for installation instructions and detailed usage. h3>To install the lightweight CLI for browsing and downloading models: h3> shell h3>pip install qai-hub-models-cli h3> h3> To install the full package for exporting, evaluating, and running demos: h3> shell h3>pip install qai_hub_models h3> h3> After installation, use the CLI: h3> shell h3>qai-hub-models models # browse the catalog h3>qai-hub-models info mobilenet_v2 # model details h3>qai-hub-models fetch mobilenet_v2 --runtime tflite --precision float # download h3> h3> For end-to-end demos, install model-specific dependencies and run: h3> shell h3>qai-hub-models install yolov7 h3>qai-hub-models demo yolov7 h3> h3>
License and pricing
h3> h3>Licensed under BSD-3-Clause. Pricing is not stated; the project is listed as open source.
Based on the project's public README and repository metadata.
