quantized int8 storage and operation

[nihui]

No description provided.

[nihui]

• int8/uint8 storage support in Mat
• Quantize Dequantize layer
• Integer convolution with bias
• ReLU6 layer (implemented as Clip)
• tensorflow quantized mobilenet model
• arm neon optimization

[yanghanxy]

This project seems like a good reference to quantize a NN model under embedded environment:
https://github.com/ARM-software/ML-KWS-for-MCU/blob/master/Deployment/Quant_guide.md

[blueardour]

Hi, I wonder if the computing of each layer is fixed point data type. As far as I know, the ristretto implements their dynamic fixed point data type by simulation. They just tranfer the data to fixed point and assgin back using float. Thus the real computing is still based on float.

[nihui]

POC implemetation #487

[nihui]

framework int8 a169cec

x86 int8 4be27a0

armv7 int8 e34aa77