Details
Presenter(s)
![Alessandro Capotondi Headshot](https://confcats-catavault.s3.amazonaws.com/CATAVault/ieeecass/master/files/styles/cc_user_photo/s3/user-pictures/23951.jpg?h=a9ff4db5&itok=3oqJxEuQ)
Display Name
Alessandro Capotondi
- Affiliation
-
AffiliationUniversità degli Studi di Modena e Reggio Emilia
- Country
Abstract
Low-precision integer arithmetic is a necessary ingredient for enabling Deep Learning inference on tiny and resource-constrained IoT edge devices. This work presents CMix-NN, a flexible open-source1 mixed low-precision (independent tensors quantization of weight and activations at 8, 4, 2 bits) inference library for low bitwidth Quantized Networks. Thanks to CMix-NN, we deploy on an STM32H7 microcontroller a set of Mobilenet family networks with the largest input resolutions (224) and highest accuracies (up to 68% Top1) when compressed with a mixed low precision technique, witnessing up to +8% accuracy improvement concerning any other previously published solutions for MCU devices.