モデルと処理速度について

Owned by Rie Sakito (崎戸 梨恵) (Unlicensed)
Last updated: Mar 27, 2024 by Tachibana Yuta (立花 裕太)
1 min read

Updated:Mar 27, 2024

質問

動作実績のあるモデルとその処理時間が知りたい。

回答

以下の表をご参照ください。

 

モデル

処理時間 (ms)　※参考値

CV22

CV25m

CV52

mobilefacenets_fix8_112_96

2.0846

4.1692

0.478

mobilenetv1_ssd_fix8_300_300

6.9175

13.835

2.393

resnet50_fix8_224_224

18.7301

37.4602

6.591

vgg19_fix8_224_224

88.0988

176.1976

28.97

vgg16_ssd_fix8_300_300

122.4476

244.8952

34.76

yolov3_fastest_fix8_320_320

1.1967

2.3934

0.675

yolov3_mnetv2_nano_fix8_320_320

1.7658

3.5316

0.824

yolov3_fastest_xl_fix8_320_320

2.5402

5.0804

1.187

yolov3_mnetv2_lite_fix8_320_320

5.0584

10.1168

1.942

tf_yolo_v3_mixed_416_416

180.2604

360.5208

47.4

tf_deeplab_v3_fix8_513_513

17.236

34.472

7.322

bodypix_fix8_481_641

25.4898

50.9796

8.135

tf_inceptionv1_fix8_224_224

6.8038

13.6076

2.254

tf_inception_v3_fix8_299_299

28.917

57.834

9.373

tf_mobilenetv1_fix8_224_224

3.2801

6.5602

1.408

tf_mobilenetv2_fix8_224_224

2.4037

4.8074

1.33

tf_mobilenetv2_ssdlite_fix8_300_300

6.0922

12.1844

3.076

tf_efficientdet_d0_fix8_512_512

79.399

158.798

63.26

onnx_efficientnet_b0_fix8_224_224

11.6885

23.377

9.56

onnx_centernet_fix8_384_384

65.2046

130.4092

18.47

onnx_fgfd_fix8_240_320

0.7421

1.4842

0.434

onnx_retinaface_fix8_640_640

4.9085

9.817

1.621

onnx_yolov3_fix8_416_416

166.9117

333.8234

45.18

onnx_yolov3_spp_fix8_416_416

167.585

335.17

45.18

onnx_yolov3_tiny_fix8_416_416

13.2943

26.5886

3.873

onnx_yolov5s_fix8_416_416

21.9484

43.8968

7.101

onnx_yolov5m_fix8_416_416

60.7822

121.5644

17.42

onnx_yolov5l_fix8_416_416

130.7734

261.5468

36.56

onnx_yolov5x_fix8_416_416

239.9791

479.9582

67.72

onnx_fairmot_fix8_160_288

26.883

53.766

8.489