Table 1 Comparison with reported in-sensor encoding neurons

From: Crossmodal sensory neurons based on high-performance flexible memristors for human-machine in-sensor computing system

Sensory signals

Crossmodal sensory

Sensory components

Flexibility

Endurance

Energy/spike

Coding-related applications

Ref.

Optical

No

1 M

No

>100

2.1–20.3 nJ

Machine vision

25

Pressure &Temperature

Yes

1 M + 1PS

No

>106

/

Object recognition

29

Optical

No

1 M

No

>104

~190 nJa

Image segmentation

57

Temperature

No

1 M

No

>103

~0.15 nJa

Edge detection

58

Physiological signals

No

2 T

No

/

~0.5 μJa

Neuromorphic bio-interface

59

Optical

No

1 M

No

>500

~32 pJa

Pattern recognition

60

Pressure & optical

Yes

1 T + 1PS

No

/

~8 nJa

/

61

Optical

No

1 T

No

>108

~0.1 nJ

Motion detection

62

Pressure &Temperature

Yes

1 M + 1PS

Yes

>1012

3.9-50 nJ

Human-machine interaction & dynamic object recognition

This work

  1. aThe energy consumption per spike is calculated approximately from the P–t, V–t and I–t curves in these reference papers, respectively.
  2. To unify the benchmark, all the sensory components in these reference papers are equivalent to three categories: memristor (M), transistor (T), and pressure sensor (PS).
Back to article page