Just addition about quantized model

pb-julian / liteface

InsightFace with TensorFlow Lite to be deployed and used in Android, iOS, embedded devices etc... for real-time face detection and recognition.

30 stars 8 forks source link

Just addition about quantized model #4

Open opiumfive opened 3 years ago

opiumfive commented 3 years ago

Hi, thanks for your tutorial, it really helped.

Just an addition to get models with smaller size to use on mobile devices, example:

tflite_convert --output_file=tf_arcface_34_v1/tf_arcface_34_v1_q.tflite --saved_model_dir=tf_arcface_34_v1 --inference_type=QUANTIZED_UINT8 --std_dev_values=128 --mean_values=128 --default_ranges_min=-6 --default_ranges_max=6 --output_format=TFLITE

kivan11 commented 2 years ago

@opiumfive hi, plz can u tell me how to complete this with flutter