Demo Team / Wildlife Camera: On Device - North America Public
Target: Sony Spresense (Cortex-M4F 156MHz) 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

No name set

Sony Spresense (Cortex-M4F 156MHz)

100 ms

1536 kB

8096 kB

Filters

Status

DSP type

Model type

View

Data set

Variant

Sort

General

F1-score

Precision

Recall

5%
grayscale-mobilenetv2-1e3
PERFORMANCE
LATENCY
461 ms of 100 ms
Exceeds target by 361 ms
RAM
351 kB of 1536 kB
ROM
648 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

Grayscale

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

0.0005 | 20 | 5%

MobileNetV2 0.35
64 | 0.2 |

5/22/2022, 10:41:35 PM

5%
grayscale-mobilenetv2-72b
PERFORMANCE
LATENCY
8007 ms of 100 ms
Exceeds target by 7907 ms
RAM
351 kB of 1536 kB
ROM
586 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

Grayscale

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

0.0005 | 20 | 5%

MobileNetV2 0.35
16 | 0.2 |

5/22/2022, 9:41:53 PM

5%
grayscale-mobilenetv2-f30
PERFORMANCE
LATENCY
455 ms of 100 ms
Exceeds target by 355 ms
RAM
297 kB of 1536 kB
ROM
285 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

Grayscale

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

0.0005 | 20 | 5%

MobileNetV2 0.1
64 | 0.5 |

5/22/2022, 9:41:50 PM

5%
grayscale-mobilenetv2-c77
PERFORMANCE
LATENCY
475 ms of 100 ms
Exceeds target by 375 ms
RAM
351 kB of 1536 kB
ROM
586 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

Grayscale

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

0.0005 | 20 | 5%

MobileNetV2 0.35
16 | 0.1

5/22/2022, 10:35:50 PM

5%
rgb-conv2d-837
PERFORMANCE
LATENCY
1492 ms of 100 ms
Exceeds target by 1392 ms
RAM
30 kB of 1536 kB
ROM
62 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

32 |
32

IMAGE

RGB

ACCURACY (KERAS)
CLASSIFICATION (KERAS)

0.0005 | 10 | 5%

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

5/22/2022, 10:25:50 PM

5%
grayscale-mobilenetv2-86e
PERFORMANCE
LATENCY
397 ms of 100 ms
Exceeds target by 297 ms
RAM
297 kB of 1536 kB
ROM
285 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

Grayscale

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

0.0005 | 20 | 5%

MobileNetV2 0.1
64 | 0.5

5/22/2022, 10:24:17 PM

5%
rgb-conv2d-b9d
PERFORMANCE
LATENCY
1113 ms of 100 ms
Exceeds target by 1013 ms
RAM
27 kB of 1536 kB
ROM
46 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

32 |
32

IMAGE

RGB

ACCURACY (KERAS)
CLASSIFICATION (KERAS)

0.0005 | 10 | 5%

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

5/22/2022, 10:02:26 PM

5%
grayscale-mobilenetv2-588
PERFORMANCE
LATENCY
3890 ms of 100 ms
Exceeds target by 3790 ms
RAM
297 kB of 1536 kB
ROM
223 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

Grayscale

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

0.0005 | 20 | 5%

MobileNetV2 0.1
16 | 0.5

5/22/2022, 10:01:03 PM

5%
grayscale-mobilenetv1-a69
PERFORMANCE
LATENCY
5598 ms of 100 ms
Exceeds target by 5498 ms
RAM
135 kB of 1536 kB
ROM
312 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

Grayscale

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

0.0005 | 20 | 5%

MobileNetV1 0.25
16 | 0.2 |

5/22/2022, 9:56:50 PM

2%
rgb-mobilenetv2-48a
PERFORMANCE
LATENCY
3504 ms of 100 ms
Exceeds target by 3404 ms
RAM
351 kB of 1536 kB
ROM
648 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

RGB

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

0.0005 | 20 | 2%

MobileNetV2 160x160 0.35
64 | 0.2

5/22/2022, 10:34:20 PM

1%
rgb-mobilenetv2-9ac
PERFORMANCE
LATENCY
3850 ms of 100 ms
Exceeds target by 3750 ms
RAM
351 kB of 1536 kB
ROM
586 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

RGB

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

0.0005 | 20 | 1%

MobileNetV2 0.35
16 | 0.2

5/22/2022, 10:11:46 PM

1%
grayscale-mobilenetv2-79f
PERFORMANCE
LATENCY
868 ms of 100 ms
Exceeds target by 768 ms
RAM
674 kB of 1536 kB
ROM
239 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

160 |
160

IMAGE

Grayscale

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

0.0005 | 20 | 1%

MobileNetV2 0.05
64 | 0.2

5/22/2022, 10:26:59 PM

1%
rgb-mobilenetv2-abc
PERFORMANCE
LATENCY
3570 ms of 100 ms
Exceeds target by 3470 ms
RAM
297 kB of 1536 kB
ROM
285 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

RGB

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

0.0005 | 20 | 1%

MobileNetV2 0.1
64 | 0.1 |

5/22/2022, 10:16:26 PM

1%
rgb-mobilenetv2-e68
PERFORMANCE
LATENCY
547 ms of 100 ms
Exceeds target by 447 ms
RAM
297 kB of 1536 kB
ROM
223 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

96 |
96

IMAGE

RGB

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

0.0005 | 20 | 1%

MobileNetV2 0.1
16 | 0.1

5/22/2022, 9:58:45 PM

1%
rgb-mobilenetv2-326
PERFORMANCE
LATENCY
736 ms of 100 ms
Exceeds target by 636 ms
RAM
674 kB of 1536 kB
ROM
239 kB of 8096 kB
DSP NN Unused
IMAGE INPUT

160 |
160

IMAGE

RGB

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

0.0005 | 20 | 1%

MobileNetV2 0.05
64 | 0.2

5/22/2022, 10:49:02 PM

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