Object Detection
PyTorch
real_time
android

YoloV5-Face: Optimized for Qualcomm Devices

YoloV5-Face-Nano is a lightweight single-class face detector based on the YoloV5 backbone (StemBlock + ShuffleV2Block). Each detected face is accompanied by 5 facial landmark predictions (left eye, right eye, nose, left mouth corner, right mouth corner). Trained on the WiderFace dataset.

This is based on the implementation of YoloV5-Face found here. 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

Due to licensing restrictions, we cannot distribute pre-exported model assets for this model. 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

See our repository for YoloV5-Face on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

  • Input resolution: 640x640
  • Model checkpoint: yolov5n-face.pt
  • Model size (float): 7.21 MB
  • Model size (w8a16): 3.68 MB
  • Number of parameters: 1.87M

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
YoloV5-Face ONNX float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 5.702 ms 1 - 187 MB NPU
YoloV5-Face ONNX float Snapdragon® 8 Elite For Galaxy Mobile 6.328 ms 3 - 186 MB NPU
YoloV5-Face ONNX float Snapdragon® X2 Elite 6.057 ms 9 - 9 MB NPU
YoloV5-Face ONNX float Snapdragon® X Elite 13.753 ms 9 - 9 MB NPU
YoloV5-Face ONNX float Snapdragon® 8 Gen 3 Mobile 7.564 ms 8 - 215 MB NPU
YoloV5-Face ONNX float Snapdragon® 8 Gen 1 Mobile 18.485 ms 8 - 231 MB NPU
YoloV5-Face ONNX float Qualcomm® Dragonwing™ IQ-8275 11.232 ms 6 - 10 MB NPU
YoloV5-Face ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 13.175 ms 6 - 9 MB NPU
YoloV5-Face ONNX float Qualcomm® QCS8450 18.485 ms 8 - 231 MB NPU
YoloV5-Face ONNX float Qualcomm® Dragonwing™ IQ-9075 14.747 ms 7 - 11 MB NPU
YoloV5-Face ONNX float Qualcomm® Dragonwing™ IQ-X7181 13.753 ms 9 - 9 MB NPU
YoloV5-Face ONNX float Qualcomm® Dragonwing™ Q-8750 6.328 ms 3 - 186 MB NPU
YoloV5-Face ONNX w8a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 5.976 ms 1 - 229 MB NPU
YoloV5-Face ONNX w8a16 Snapdragon® 8 Elite For Galaxy Mobile 6.308 ms 2 - 223 MB NPU
YoloV5-Face ONNX w8a16 Snapdragon® X2 Elite 6.032 ms 4 - 4 MB NPU
YoloV5-Face ONNX w8a16 Snapdragon® X Elite 14.335 ms 4 - 4 MB NPU
YoloV5-Face ONNX w8a16 Snapdragon® 8 Gen 3 Mobile 7.735 ms 4 - 253 MB NPU
YoloV5-Face ONNX w8a16 Snapdragon® 8 Gen 1 Mobile 15.031 ms 4 - 254 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® Dragonwing™ QCS6490 28.842 ms 3 - 6 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® Dragonwing™ IQ-8275 10.625 ms 3 - 7 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 13.669 ms 3 - 6 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® QCS8450 15.031 ms 4 - 254 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® Dragonwing™ IQ-9075 14.601 ms 4 - 8 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® Dragonwing™ IQ-X7181 14.335 ms 4 - 4 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® Dragonwing™ Q-6690 56.563 ms 4 - 245 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® Dragonwing™ Q-7790 14.001 ms 4 - 229 MB NPU
YoloV5-Face ONNX w8a16 Qualcomm® Dragonwing™ Q-8750 6.308 ms 2 - 223 MB NPU
YoloV5-Face ONNX w8a16 Snapdragon® 7 Gen 4 Mobile 14.001 ms 4 - 229 MB NPU
YoloV5-Face QNN_DLC float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 1.649 ms 2 - 60 MB NPU
YoloV5-Face QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 2.172 ms 0 - 58 MB NPU
YoloV5-Face QNN_DLC float Snapdragon® X2 Elite 2.313 ms 5 - 5 MB NPU
YoloV5-Face QNN_DLC float Snapdragon® X Elite 4.551 ms 5 - 5 MB NPU
YoloV5-Face QNN_DLC float Snapdragon® 8 Gen 3 Mobile 2.893 ms 0 - 76 MB NPU
YoloV5-Face QNN_DLC float Snapdragon® 8 Gen 1 Mobile 7.951 ms 5 - 100 MB NPU
YoloV5-Face QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 4.934 ms 5 - 14 MB NPU
YoloV5-Face QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 3.98 ms 5 - 60 MB NPU
YoloV5-Face QNN_DLC float Qualcomm® QCS8450 7.951 ms 5 - 100 MB NPU
YoloV5-Face QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 5.063 ms 7 - 15 MB NPU
YoloV5-Face QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 4.551 ms 5 - 5 MB NPU
YoloV5-Face QNN_DLC float Qualcomm® Dragonwing™ Q-8750 2.172 ms 0 - 58 MB NPU
YoloV5-Face QNN_DLC float Qualcomm® SA7255P 12.581 ms 1 - 57 MB NPU
YoloV5-Face QNN_DLC float Qualcomm® SA8295P 8.817 ms 1 - 71 MB NPU
YoloV5-Face QNN_DLC w8a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 1.618 ms 2 - 195 MB NPU
YoloV5-Face QNN_DLC w8a16 Snapdragon® 8 Elite For Galaxy Mobile 2.05 ms 0 - 191 MB NPU
YoloV5-Face QNN_DLC w8a16 Snapdragon® X2 Elite 2.083 ms 2 - 2 MB NPU
YoloV5-Face QNN_DLC w8a16 Snapdragon® X Elite 4.956 ms 2 - 2 MB NPU
YoloV5-Face QNN_DLC w8a16 Snapdragon® 8 Gen 3 Mobile 3.03 ms 2 - 107 MB NPU
YoloV5-Face QNN_DLC w8a16 Snapdragon® 8 Gen 1 Mobile 5.109 ms 2 - 107 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® Dragonwing™ QCS6490 11.497 ms 1 - 6 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-8275 4.356 ms 2 - 8 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 4.445 ms 2 - 18 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® SA8775P 5.136 ms 2 - 83 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® SA8650P 5.136 ms 2 - 83 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® SA8255P 5.136 ms 2 - 83 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® QCS8450 5.109 ms 2 - 107 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-9075 4.967 ms 3 - 8 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-X7181 4.956 ms 2 - 2 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-6690 17.795 ms 2 - 192 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-7790 4.9 ms 2 - 200 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-8750 2.05 ms 0 - 191 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® SA7255P 8.36 ms 1 - 81 MB NPU
YoloV5-Face QNN_DLC w8a16 Qualcomm® SA8295P 5.638 ms 0 - 78 MB NPU
YoloV5-Face QNN_DLC w8a16 Snapdragon® 7 Gen 4 Mobile 4.9 ms 2 - 200 MB NPU
YoloV5-Face TFLITE float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 1.66 ms 0 - 60 MB NPU
YoloV5-Face TFLITE float Snapdragon® 8 Elite For Galaxy Mobile 2.174 ms 0 - 59 MB NPU
YoloV5-Face TFLITE float Snapdragon® 8 Gen 3 Mobile 2.871 ms 0 - 79 MB NPU
YoloV5-Face TFLITE float Snapdragon® 8 Gen 1 Mobile 7.937 ms 1 - 101 MB NPU
YoloV5-Face TFLITE float Qualcomm® Dragonwing™ IQ-8275 5.008 ms 1 - 16 MB NPU
YoloV5-Face TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 3.958 ms 0 - 2 MB NPU
YoloV5-Face TFLITE float Qualcomm® SA8775P 5.402 ms 2 - 65 MB NPU
YoloV5-Face TFLITE float Qualcomm® SA8650P 5.402 ms 2 - 65 MB NPU
YoloV5-Face TFLITE float Qualcomm® SA8255P 5.402 ms 2 - 65 MB NPU
YoloV5-Face TFLITE float Qualcomm® QCS8450 7.937 ms 1 - 101 MB NPU
YoloV5-Face TFLITE float Qualcomm® Dragonwing™ IQ-9075 5.159 ms 1 - 15 MB NPU
YoloV5-Face TFLITE float Qualcomm® Dragonwing™ Q-8750 2.174 ms 0 - 59 MB NPU
YoloV5-Face TFLITE float Qualcomm® SA7255P 12.605 ms 2 - 63 MB NPU
YoloV5-Face TFLITE float Qualcomm® SA8295P 8.887 ms 2 - 73 MB NPU

License

  • The license for the original implementation of YoloV5-Face can be found here.

References

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

Paper for qualcomm/YoloV5-Face