Table 3 Performance comparison of the proposed P2M-compatible models with state-of-the-art deep CNNs on VWW dataset.

From: A processing-in-pixel-in-memory paradigm for resource-constrained TinyML applications

Authors

Description

Model architecture

Test accuracy (%)

Saha et al.33

RNNPooling

MobileNetV2

89.65

Han et al.45

ProxylessNAS

Non-standard architecture

90.27

Banbury et al.38

Differentiable NAS

MobileNet-V2

88.75

Zhoue et al.46

Analog compute-in-memory

MobileNet-V2

85.7

This work

P2M

MobileNet-V2

89.90

Back to article page