search

jiangxiluning / facenet_mtcnn_to_mobile

convert facenet and mtcnn models from tensorflow to tensorflow lite and coreml （使用 TFLite 将 FaceNet 和 MTCNN 移植到移动端）
MIT License
197 stars 49 forks source link

Quantized inference #11

Open rushidesai opened 5 years ago

rushidesai commented 5 years ago

I want to convert this model to run on Google's Edge TPU (Coral), but it needs an quantized model. Not sure what mean/std_dev values to use. Also Abs does not have a quantized form implemented yet. Wondering if there is a workaround for that.