wang / waste sorting Public
Target: Cortex-M7 216MHz Clone this project
Experiments EON Tuner
The EON Tuner helps you quickly run hyper-parameter sweeps that explore different pre-processing + model architectures optimized for your defined objectives. Clone this project to use the EON Tuner.

Target

waste

Cortex-M7 216MHz

100 ms

340 kB

1024 kB

Filters

Status

DSP type

Model type

View

Data set

Variant

Sort

General

F1-score

Precision

Recall

Run #1 - waste (Apr 16 2024, 13:40:29, finished)
72%
rgb-mobilenetv2-55e
PERFORMANCE
LATENCY
38 ms of 100 ms
RAM
275 kB of 340 kB
ROM
234 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

RGB

ACCURACY (KERAS-TRANSFER-IMAGE)
TRANSFER LEARNING (IMAGES)

0.0005 | 20 | 72%

MobileNetV2 0.05
64 | 0.1

4/16/2024, 2:11:25 PM

70%
rgb-mobilenetv2-068
PERFORMANCE
LATENCY
80 ms of 100 ms
RAM
339 kB of 340 kB
ROM
647 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

RGB

ACCURACY (KERAS-TRANSFER-IMAGE)
TRANSFER LEARNING (IMAGES)

0.0005 | 20 | 70%

MobileNetV2 0.35
64 | 0.5

4/16/2024, 2:00:02 PM

67%
grayscale-mobilenetv1-b6a
PERFORMANCE
LATENCY
116 ms of 100 ms
Exceeds target by 16 ms
RAM
258 kB of 340 kB
ROM
307 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

160 |
160

IMAGE

Grayscale

ACCURACY (KERAS-TRANSFER-IMAGE)
TRANSFER LEARNING (IMAGES)

0.0005 | 20 | 67%

MobileNetV1 0.25
16 | 0.1 |

4/16/2024, 1:59:45 PM

66%
rgb-mobilenetv2-261
PERFORMANCE
LATENCY
23 ms of 100 ms
RAM
285 kB of 340 kB
ROM
222 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

RGB

ACCURACY (KERAS-TRANSFER-IMAGE)
TRANSFER LEARNING (IMAGES)

0.0005 | 20 | 66%

MobileNetV2 0.1
16 | 0.5

4/16/2024, 1:57:27 PM

66%
rgb-mobilenetv1-45b
PERFORMANCE
LATENCY
135 ms of 100 ms
Exceeds target by 35 ms
RAM
258 kB of 340 kB
ROM
307 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

160 |
160

IMAGE

RGB

ACCURACY (KERAS-TRANSFER-IMAGE)
TRANSFER LEARNING (IMAGES)

0.0005 | 20 | 66%

MobileNetV1 0.25
16 | 0.5 |

4/16/2024, 2:02:35 PM

62%
rgb-mobilenetv1-d98
PERFORMANCE
LATENCY
42 ms of 100 ms
RAM
123 kB of 340 kB
ROM
110 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

160 |
160

IMAGE

RGB

ACCURACY (KERAS-TRANSFER-IMAGE)
TRANSFER LEARNING (IMAGES)

0.0005 | 20 | 62%

MobileNetV1 0.1
64 | 0.1

4/16/2024, 2:21:11 PM

61%
rgb-mobilenetv2-99a
PERFORMANCE
LATENCY
102 ms of 100 ms
Exceeds target by 2 ms
RAM
672 kB of 340 kB
Exceeds target by 332 kB
ROM
223 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

160 |
160

IMAGE

RGB

ACCURACY (KERAS-TRANSFER-IMAGE)
TRANSFER LEARNING (IMAGES)

0.0005 | 20 | 61%

MobileNetV2 0.1
16 | 0.1 |

4/16/2024, 2:07:19 PM

58%
grayscale-mobilenetv2-5d1
PERFORMANCE
LATENCY
35 ms of 100 ms
RAM
275 kB of 340 kB
ROM
234 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

Grayscale

ACCURACY (KERAS-TRANSFER-IMAGE)
TRANSFER LEARNING (IMAGES)

0.0005 | 20 | 58%

MobileNetV2 0.05
64 | 0.1 |

4/16/2024, 1:55:39 PM

39%
grayscale-conv2d-279
PERFORMANCE
LATENCY
16 ms of 100 ms
RAM
49 kB of 340 kB
ROM
60 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

64 |
64

IMAGE

Grayscale

ACCURACY (KERAS)
CLASSIFICATION

0.0005 | 10 | 39%

Type Filters Kernel Rate
conv2d 8 3 -
conv2d 16 3 -
conv2d 32 3 -
conv2d 64 3 -
dropout - - 0.5

4/16/2024, 1:49:21 PM

30%
grayscale-conv2d-85b
PERFORMANCE
LATENCY
6 ms of 100 ms
RAM
19 kB of 340 kB
ROM
57 kB of 1024 kB
DSP NN Unused
IMAGE INPUT

32 |
32

IMAGE

Grayscale

ACCURACY (KERAS)
CLASSIFICATION

0.0005 | 10 | 30%

Type Filters Kernel Rate
conv2d 8 3 -
conv2d 16 3 -
conv2d 32 3 -
conv2d 64 3 -
dropout - - 0.5

4/16/2024, 1:51:20 PM

Help & documentation Get help on the forum Send feedback Contact sales
Join us
Terms of Service Privacy policy