Lightweight-Face-Detection: Optimized for Qualcomm Devices

A small and accurate model for detecting bounding boxes for faces in images. This model's architecture was developed by Qualcomm. The model was trained by Qualcomm on a proprietary dataset of faces, but can be used on any image.

This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.

Getting Started

There are two ways to deploy this model on your device:

Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

Runtime Precision Chipset SDK Versions Download
ONNX float Universal QAIRT 2.42, ONNX Runtime 1.24.1 Download
ONNX w8a16 Universal QAIRT 2.42, ONNX Runtime 1.24.1 Download
ONNX w8a16_mixed_int16 Universal QAIRT 2.42, ONNX Runtime 1.24.1 Download
ONNX w8a8 Universal QAIRT 2.42, ONNX Runtime 1.24.1 Download
QNN_DLC float Universal QAIRT 2.43 Download
QNN_DLC w8a16 Universal QAIRT 2.43 Download
QNN_DLC w8a16_mixed_int16 Universal QAIRT 2.43 Download
QNN_DLC w8a8 Universal QAIRT 2.43 Download
TFLITE float Universal QAIRT 2.43, TFLite 2.17.0 Download
TFLITE w8a8 Universal QAIRT 2.43, TFLite 2.17.0 Download

For more device-specific assets and performance metrics, visit Lightweight-Face-Detection on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:

  • Custom weights (e.g., fine-tuned checkpoints)
  • Custom input shapes
  • Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for Lightweight-Face-Detection on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

  • Model checkpoint: qfd360_sl_model.pt
  • Inference latency: RealTime
  • Input resolution: 480x640
  • Number of parameters: 878K
  • Model size (float): 3.37 MB
  • Model size (w8a8): 965 KB
  • Model size (w8a16): 1.09 MB

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
Lightweight-Face-Detection ONNX float Snapdragon® 8 Elite Gen 5 Mobile 1.076 ms 0 - 148 MB NPU
Lightweight-Face-Detection ONNX float Snapdragon® X2 Elite 1.153 ms 0 - 0 MB NPU
Lightweight-Face-Detection ONNX float Snapdragon® X Elite 2.217 ms 1 - 1 MB NPU
Lightweight-Face-Detection ONNX float Snapdragon® 8 Gen 3 Mobile 1.087 ms 0 - 162 MB NPU
Lightweight-Face-Detection ONNX float Qualcomm® QCS8550 (Proxy) 1.971 ms 1 - 3 MB NPU
Lightweight-Face-Detection ONNX float Qualcomm® QCS9075 2.497 ms 1 - 4 MB NPU
Lightweight-Face-Detection ONNX float Snapdragon® 8 Elite For Galaxy Mobile 0.895 ms 0 - 140 MB NPU
Lightweight-Face-Detection ONNX w8a16 Snapdragon® 8 Elite Gen 5 Mobile 0.831 ms 0 - 146 MB NPU
Lightweight-Face-Detection ONNX w8a16 Snapdragon® X2 Elite 0.907 ms 0 - 0 MB NPU
Lightweight-Face-Detection ONNX w8a16 Snapdragon® X Elite 1.959 ms 1 - 1 MB NPU
Lightweight-Face-Detection ONNX w8a16 Snapdragon® 8 Gen 3 Mobile 0.987 ms 0 - 156 MB NPU
Lightweight-Face-Detection ONNX w8a16 Qualcomm® QCS6490 88.233 ms 29 - 31 MB CPU
Lightweight-Face-Detection ONNX w8a16 Qualcomm® QCS8550 (Proxy) 1.745 ms 0 - 12 MB NPU
Lightweight-Face-Detection ONNX w8a16 Qualcomm® QCS9075 2.086 ms 0 - 3 MB NPU
Lightweight-Face-Detection ONNX w8a16 Qualcomm® QCM6690 44.753 ms 28 - 35 MB CPU
Lightweight-Face-Detection ONNX w8a16 Snapdragon® 8 Elite For Galaxy Mobile 0.851 ms 0 - 147 MB NPU
Lightweight-Face-Detection ONNX w8a16 Snapdragon® 7 Gen 4 Mobile 34.156 ms 29 - 37 MB CPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Snapdragon® 8 Elite Gen 5 Mobile 0.855 ms 0 - 147 MB NPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Snapdragon® X2 Elite 0.908 ms 0 - 0 MB NPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Snapdragon® X Elite 1.951 ms 1 - 1 MB NPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Snapdragon® 8 Gen 3 Mobile 1.035 ms 0 - 158 MB NPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Qualcomm® QCS6490 88.112 ms 28 - 31 MB CPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Qualcomm® QCS8550 (Proxy) 1.753 ms 0 - 102 MB NPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Qualcomm® QCS9075 2.143 ms 0 - 3 MB NPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Qualcomm® QCM6690 44.759 ms 26 - 34 MB CPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Snapdragon® 8 Elite For Galaxy Mobile 0.854 ms 0 - 147 MB NPU
Lightweight-Face-Detection ONNX w8a16_mixed_int16 Snapdragon® 7 Gen 4 Mobile 34.23 ms 28 - 35 MB CPU
Lightweight-Face-Detection ONNX w8a8 Snapdragon® 8 Elite Gen 5 Mobile 0.22 ms 0 - 145 MB NPU
Lightweight-Face-Detection ONNX w8a8 Snapdragon® X2 Elite 0.25 ms 0 - 0 MB NPU
Lightweight-Face-Detection ONNX w8a8 Snapdragon® X Elite 0.558 ms 0 - 0 MB NPU
Lightweight-Face-Detection ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 0.303 ms 0 - 153 MB NPU
Lightweight-Face-Detection ONNX w8a8 Qualcomm® QCS6490 17.106 ms 12 - 17 MB CPU
Lightweight-Face-Detection ONNX w8a8 Qualcomm® QCS8550 (Proxy) 0.463 ms 0 - 85 MB NPU
Lightweight-Face-Detection ONNX w8a8 Qualcomm® QCS9075 0.626 ms 0 - 3 MB NPU
Lightweight-Face-Detection ONNX w8a8 Qualcomm® QCM6690 12.875 ms 11 - 19 MB CPU
Lightweight-Face-Detection ONNX w8a8 Snapdragon® 8 Elite For Galaxy Mobile 0.245 ms 0 - 140 MB NPU
Lightweight-Face-Detection ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 9.66 ms 13 - 21 MB CPU
Lightweight-Face-Detection QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 1.076 ms 1 - 33 MB NPU
Lightweight-Face-Detection QNN_DLC float Snapdragon® X2 Elite 1.319 ms 1 - 1 MB NPU
Lightweight-Face-Detection QNN_DLC float Snapdragon® X Elite 2.801 ms 1 - 1 MB NPU
Lightweight-Face-Detection QNN_DLC float Snapdragon® 8 Gen 3 Mobile 1.596 ms 0 - 42 MB NPU
Lightweight-Face-Detection QNN_DLC float Qualcomm® QCS8275 (Proxy) 6.652 ms 1 - 29 MB NPU
Lightweight-Face-Detection QNN_DLC float Qualcomm® QCS8550 (Proxy) 2.543 ms 1 - 2 MB NPU
Lightweight-Face-Detection QNN_DLC float Qualcomm® SA8775P 3.251 ms 0 - 30 MB NPU
Lightweight-Face-Detection QNN_DLC float Qualcomm® QCS9075 3.701 ms 1 - 4 MB NPU
Lightweight-Face-Detection QNN_DLC float Qualcomm® QCS8450 (Proxy) 4.296 ms 0 - 43 MB NPU
Lightweight-Face-Detection QNN_DLC float Qualcomm® SA7255P 6.652 ms 1 - 29 MB NPU
Lightweight-Face-Detection QNN_DLC float Qualcomm® SA8295P 3.727 ms 0 - 26 MB NPU
Lightweight-Face-Detection QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 1.222 ms 0 - 28 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Snapdragon® 8 Elite Gen 5 Mobile 0.706 ms 1 - 142 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Snapdragon® X2 Elite 0.939 ms 1 - 1 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Snapdragon® X Elite 1.967 ms 1 - 1 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Snapdragon® 8 Gen 3 Mobile 0.957 ms 0 - 153 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® QCS6490 6.443 ms 0 - 2 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® QCS8275 (Proxy) 2.963 ms 1 - 139 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® QCS8550 (Proxy) 1.732 ms 1 - 2 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® SA8775P 1.973 ms 1 - 143 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® QCS9075 2.173 ms 1 - 3 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® QCM6690 14.61 ms 1 - 140 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® QCS8450 (Proxy) 2.898 ms 0 - 154 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® SA7255P 2.963 ms 1 - 139 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Qualcomm® SA8295P 2.278 ms 0 - 137 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Snapdragon® 8 Elite For Galaxy Mobile 0.774 ms 0 - 139 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16 Snapdragon® 7 Gen 4 Mobile 2.584 ms 1 - 139 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Snapdragon® 8 Elite Gen 5 Mobile 0.736 ms 1 - 142 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Snapdragon® X2 Elite 0.962 ms 1 - 1 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Snapdragon® X Elite 1.988 ms 1 - 1 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Snapdragon® 8 Gen 3 Mobile 0.961 ms 0 - 153 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Qualcomm® QCS8275 (Proxy) 3.188 ms 1 - 139 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Qualcomm® QCS8550 (Proxy) 1.771 ms 1 - 23 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Qualcomm® SA8775P 8.537 ms 1 - 139 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Qualcomm® QCS9075 2.402 ms 2 - 4 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Qualcomm® QCM6690 13.212 ms 1 - 141 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Qualcomm® SA7255P 3.188 ms 1 - 139 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Snapdragon® 8 Elite For Galaxy Mobile 0.792 ms 1 - 143 MB NPU
Lightweight-Face-Detection QNN_DLC w8a16_mixed_int16 Snapdragon® 7 Gen 4 Mobile 2.543 ms 0 - 139 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 Mobile 0.193 ms 0 - 27 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Snapdragon® X2 Elite 0.303 ms 0 - 0 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Snapdragon® X Elite 0.531 ms 0 - 0 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 0.282 ms 0 - 34 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® QCS6490 1.449 ms 0 - 2 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® QCS8275 (Proxy) 1.09 ms 0 - 25 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® QCS8550 (Proxy) 0.411 ms 0 - 4 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® SA8775P 0.607 ms 0 - 27 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® QCS9075 0.556 ms 0 - 2 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® QCM6690 2.808 ms 0 - 137 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® QCS8450 (Proxy) 0.495 ms 0 - 36 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® SA7255P 1.09 ms 0 - 25 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Qualcomm® SA8295P 0.833 ms 0 - 22 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Snapdragon® 8 Elite For Galaxy Mobile 0.23 ms 0 - 28 MB NPU
Lightweight-Face-Detection QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 0.479 ms 0 - 137 MB NPU
Lightweight-Face-Detection TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 1.069 ms 0 - 34 MB NPU
Lightweight-Face-Detection TFLITE float Snapdragon® 8 Gen 3 Mobile 1.61 ms 0 - 44 MB NPU
Lightweight-Face-Detection TFLITE float Qualcomm® QCS8275 (Proxy) 6.573 ms 0 - 29 MB NPU
Lightweight-Face-Detection TFLITE float Qualcomm® QCS8550 (Proxy) 2.575 ms 0 - 1 MB NPU
Lightweight-Face-Detection TFLITE float Qualcomm® SA8775P 3.266 ms 0 - 31 MB NPU
Lightweight-Face-Detection TFLITE float Qualcomm® QCS9075 3.739 ms 0 - 5 MB NPU
Lightweight-Face-Detection TFLITE float Qualcomm® QCS8450 (Proxy) 4.301 ms 0 - 44 MB NPU
Lightweight-Face-Detection TFLITE float Qualcomm® SA7255P 6.573 ms 0 - 29 MB NPU
Lightweight-Face-Detection TFLITE float Qualcomm® SA8295P 3.735 ms 0 - 27 MB NPU
Lightweight-Face-Detection TFLITE float Snapdragon® 8 Elite For Galaxy Mobile 1.23 ms 0 - 34 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Snapdragon® 8 Elite Gen 5 Mobile 0.194 ms 0 - 26 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 0.282 ms 0 - 34 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® QCS6490 1.366 ms 0 - 3 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® QCS8275 (Proxy) 1.121 ms 0 - 24 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® QCS8550 (Proxy) 0.412 ms 0 - 1 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® SA8775P 0.627 ms 0 - 26 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® QCS9075 0.575 ms 0 - 3 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® QCM6690 2.814 ms 0 - 137 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® QCS8450 (Proxy) 0.502 ms 0 - 36 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® SA7255P 1.121 ms 0 - 24 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Qualcomm® SA8295P 0.861 ms 0 - 22 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Snapdragon® 8 Elite For Galaxy Mobile 0.238 ms 0 - 28 MB NPU
Lightweight-Face-Detection TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 0.482 ms 0 - 140 MB NPU

License

  • The license for the original implementation of Lightweight-Face-Detection can be found here.

Community

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support