Skip to content

fanz-xlnx/AI-Model-Zoo

BranchesTags
 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
images
images
 
 
LICENSE
LICENSE
 
 
README.md
README.md
 
 
get_model.sh
get_model.sh
 
 

Repository files navigation

Vitis AI Model Zoo

Introduction

This repository includes optimized deep learning models to speed up the deployment of deep learning inference on Xilinx™ platforms. These models cover different applications, including but not limited to ADAS/AD, video surveillance, robotics, data center, etc. You can get started with these free pre-trained models to enjoy the benefits of deep learning acceleration.

Model Information

The following table includes comprehensive information about each model, including application, framework, training and validation dataset, backbone, input size, computation as well as float and quantized precision.

Click here to view details
No. Application Model Name Framework Backbone Input Size OPS per image Training Set Val Set Float (Top1, Top5)/ mAP/mIoU Quantized (Top1, Top5)/mAP/mIoU
1 Image Classification resnet50 cf_resnet50_imagenet_224_224_7.7G_1.1 caffe resnet50 224*224 7.7G ImageNet Train ImageNet Validataion 0.7444/0.9185 0.7334/0.9131
2 Image Classifiction resnet18 cf_resnet18_imagenet_224_224_3.65G_1.1 caffe resnet18 224*224 3.65G ImageNet Train ImageNet Validataion 0.6832/0.8848 66.94%/88.25%
3 Image Classification Inception_v1 cf_inceptionv1_imagenet_224_224_3.16G_1.1 caffe inception_v1 224*224 3.16G ImageNet Train ImageNet Validataion 0.7030/0.8971 0.6984/0.8942
4 Image Classification Inception_v2 cf_inceptionv2_imagenet_224_224_4G_1.1 caffe bn-inception 224*224 4G ImageNet Train ImageNet Validataion 0.7275/0.9111 0.7168/0.9029
5 Image Classification Inception_v3 cf_inceptionv3_imagenet_299_299_11.4G_1.1 caffe inception_v3 299*299 11.4G ImageNet Train ImageNet Validataion 0.7701/0.9329 0.7626/0.9303
6 Image Classification Inception_v4 cf_inceptionv4_imagenet_299_299_24.5G_1.1 caffe inception_v3 299*299 24.5G ImageNet Train ImageNet Validataion 0.7958/0.9470 0.7898/0.9445
7 Image Classification mobileNet_v2 cf_mobilenetv2_imagenet_224_224_0.59G_1.1 caffe MobileNet_v2 224*224 608M ImageNet Train ImageNet Validataion 0.6475/0.8609 0.6354/0.8506
8 Image Classifiction SqueezeNet cf_squeeze_imagenet_227_227_0.76G_1.1 caffe squeezenet 227*227 0.76G ImageNet Train ImageNet Validataion 0.5438/0.7813 0.5026/0.7658
9 ADAS Pedstrain Detection ssd_pedestrain_pruned_0.97 cf_ssdpedestrian_coco_360_640_0.97_5.9G_1.1 caffe VGG-bn-16 360*640 5.9G coco2014_train_person and crowndhuman coco2014_val_person 0.5903 0.5876
10 Object Detection refinedet_pruned_0.8 cf_refinedet_coco_360_480_0.8_25G_1.1 caffe VGG-bn-16 360*480 25G coco2014_train_person coco2014_val_person 0.6794 0.6780
11 Object Detection refinedet_pruned_0.92 cf_refinedet_coco_360_480_0.92_10.10G_1.1 caffe VGG-bn-16 360*480 10.10G coco2014_train_person coco2014_val_person 0.6489 0.6486
12 Object Detection refinedet_pruned_0.96 cf_refinedet_coco_360_480_0.96_5.08G_1.1 caffe VGG-bn-16 360*480 5.08G coco2014_train_person coco2014_val_person 0.6120 0.6113
13 ADAS Vehicle Detection ssd_adas_pruned_0.95 cf_ssdadas_bdd_360_480_0.95_6.3G_1.1 caffe VGG-16 360*480 6.3G bdd100k + private data bdd100k + private data 0.4207 0.4200
14 Traffic Detection ssd_traffic_pruned_0.9 cf_ssdtraffic_360_480_0.9_11.6G_1.1 caffe VGG-16 360*480 11.6G private data private data 0.5982 0.5921
15 ADAS Lane Detection VPGnet_pruned_0.99 cf_VPGnet_caltechlane_480_640_0.99_2.5G_1.1 caffe VGG 480*640 2.5G caltech-lanes-train-dataset caltech lane 0.8864(F1-score) 0.8882(F1-score)
16 Object Detection ssd_mobilnet_v2 cf_ssdmobilenetv2_bdd_360_480_6.57G_1.1 caffe MobileNet_v2 360*480 6.57G bdd100k train bdd100k val 0.3052 0.2752
17 ADAS Segmentation FPN cf_fpn_cityscapes_256_512_8.9G_1.1 caffe Google_v1_BN 256*512 8.9G Cityscapes gtFineTrain(2975) Cityscapes Val(500) 0.5669 0.5662
18 Pose Estimation SP-net cf_SPnet_aichallenger_224_128_0.54G_1.1 caffe Google_v1_BN 128*224 548.6M ai_challenger ai_challenger 0.9000(PCKh0.5) 0.8964(PCKh0.5)
19 Pose Estimation Openpose_pruned_0.3 cf_openpose_aichallenger_368_368_0.3_189.7G_1.1 caffe VGG 368*368 49.88G ai_challenger ai_challenger 0.4507(OKs) 0.4422(Oks)
20 Face Detection densebox_320_320 cf_densebox_wider_320_320_0.49G_1.1 caffe VGG-16 320*320 0.49G wider_face FDDB 0.8833 0.8791
21 Face Detection densebox_360_640 cf_densebox_wider_360_640_1.11G_1.1 caffe VGG-16 360*640 1.11G wider_face FDDB 0.8931 0.8925
22 Face Recognition face_landmark cf_landmark_celeba_96_72_0.14G_1.1 caffe lenet 96*72 0.14G celebA processed helen 0.1952(L2 loss) 0.1972(L2 loss)
23 Re-identification reid cf_reid_market1501_160_80_0.95G_1.1 caffe resnet18 160*80 0.95G Market1501+CUHK03 Market1501 0.7800 0.7790
24 Detection+Segmentation multi-task cf_multitask_bdd_288_512_14.8G_1.1 caffe ssd 288*512 14.8G BDD100K+Cityscapes BDD100K+Cityscapes 0.2228(Det) 0.4088(Seg) 0.2202(Det) 0.4058(Seg)
25 Object Detection yolov3_bdd dk_yolov3_bdd_288_512_53.7G_1.1 darknet darknet-53 288*512 53.7G bdd100k bdd100k 0.5058 0.4914
26 Object Detection yolov3_adas_prune_0.9 dk_yolov3_cityscapes_256_512_0.9_5.46G_1.1 darknet darknet-53 256*512 5.46G cityscape train cityscape val 0.5520 0.5300
27 Object Detection yolov3_voc dk_yolov3_voc_416_416_65.42G_1.1 darknet darknet-53 416*416 65.42G voc07+12_trainval voc07_test 0.8240(MaxIntegral) 0.8150(MaxIntegral)
28 Object Detection yolov2_voc dk_yolov2_voc_448_448_34G_1.1 darknet darknet-19 448*448 34G voc07+12_trainval voc07_test 0.7845(MaxIntegral) 0.7739(MaxIntegral)
29 Object Detection yolov2_voc_pruned_0.66 dk_yolov2_voc_448_448_0.66_11.56G_1.1 darknet darknet-19 448*448 11.56G voc07+12_trainval voc07_test 0.7700(MaxIntegral) 0.7600(MaxIntegral)
30 Object Detection yolov2_voc_pruned_0.71 dk_yolov2_voc_448_448_0.71_9.86G_1.1 darknet darknet-19 448*448 9.86G voc07+12_trainval voc07_test 0.7670(MaxIntegral) 0.7530(MaxIntegral)
31 Object Detection yolov2_voc_pruned_0.77 dk_yolov2_voc_448_448_0.77_7.82G_1.1 darknet darknet-19 448*448 7.82G voc07+12_trainval voc07_test 0.7576(MaxIntegral) 0.7460(MaxIntegral)
32 Image Classifiction Inception_resnet_v2 tf_inceptionresnetv2_imagenet_299_299_26.35G_1.1 tensorflow inception 299*299 26.35G ImageNet Train ImageNet Validataion 0.8037 0.7946
33 Image Classifiction Inception_v1 tf_inceptionv1_imagenet_224_224_3G_1.1 tensorflow inception 224*224 3G ImageNet Train ImageNet Validataion 0.6976 0.6794
34 Image Classifiction Inception_v3 tf_inceptionv3_imagenet_299_299_11.45G_1.1 tensorflow inception 299*299 11.45G ImageNet Train ImageNet Validataion 0.7798 0.7607
35 Image Classifiction Inception_v4 tf_inceptionv4_imagenet_299_299_24.55G_1.1 tensorflow inception 299*299 24.55G ImageNet Train ImageNet Validataion 0.8018 0.7928
36 Image Classifiction Mobilenet_v1 tf_mobilenetv1_0.25_imagenet_128_128_27.15M_1.1 tensorflow mobilenet 128*128 27.15M ImageNet Train ImageNet Validataion 0.4144 0.3464
37 Image Classifiction Mobilenet_v1 tf_mobilenetv1_0.5_imagenet_160_160_150.07M_1.1 tensorflow mobilenet 160*160 150.07M ImageNet Train ImageNet Validataion 0.5903 0.5195
38 Image Classifiction Mobilenet_v1 tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1 tensorflow mobilenet 224*224 1.14G ImageNet Train ImageNet Validataion 0.7102 0.6779
39 Image Classifiction Mobilenet_v2 tf_mobilenetv2_1.0_imagenet_224_224_0.59G_1.1 tensorflow mobilenet 224*224 0.59G ImageNet Train ImageNet Validataion 0.7013 0.6767
40 Image Classifiction Mobilenet_v2 tf_mobilenetv2_1.4_imagenet_224_224_1.16G_1.1 tensorflow mobilenet 224*224 1.16G ImageNet Train ImageNet Validataion 0.7411 0.7194
41 Image Classifiction resnet_v1_50 tf_resnetv1_50_imagenet_224_224_6.97G_1.1 tensorflow resnetv1 224*224 6.97G ImageNet Train ImageNet Validataion 0.7520 0.7423
42 Image Classifiction resnet_v1_101 tf_resnetv1_101_imagenet_224_224_14.4G_1.1 tensorflow resnetv1 224*224 14.4G ImageNet Train ImageNet Validataion 0.7640 0.7417
43 Image Classifiction resnet_v1_152 tf_resnetv1_152_imagenet_224_224_21.83G_1.1 tensorflow resnetv1 224*224 21.83G ImageNet Train ImageNet Validataion 0.7681 0.7463
44 Image Classifiction vgg_16 tf_vgg16_imagenet_224_224_30.96G_1.1 tensorflow vgg 224*224 30.96G ImageNet Train ImageNet Validataion 0.7089 0.7069
45 Image Classifiction vgg_19 tf_vgg19_imagenet_224_224_39.28G_1.1 tensorflow vgg 224*224 39.28G ImageNet Train ImageNet Validataion 0.7100 0.7026
46 Object Detection ssd_mobilenet_v1 tf_ssdmobilenetv1_coco_300_300_2.47G_1.1 tensorflow mobilenet 300*300 2.47G coco2017 coco2014 minival 0.2080 0.1960
47 Object Detection ssd_mobilenet_v2 tf_ssdmobilenetv2_coco_300_300_3.75G_1.1 tensorflow mobilenet 300*300 3.75G coco2017 coco2014 minival 0.2150 0.2030
48 Object Detection ssd_resnet50_v1_fpn tf_ssdresnet50v1_fpn_coco_640_640_178.4G_1.1 tensorflow resnet50 300*300 178.4G coco2017 coco2014 minival 0.3010 0.2900
49 Object Detection yolov3_voc tf_yolov3_voc_416_416_65.63G_1.1 tensorflow darknet-53 416*416 65.63G voc07+12_trainval voc07_test 0.7846 0.7744
50 Object Detection mlperf_ssd_resnet34 tf_mlperf_resnet34_coco_1200_1200_433G_1.1 tensorflow resnet34 1200*1200 433G coco2017 coco2017 0.2250 0.2110

Naming Rules

Model name: F_M_D_H_W_(P)_C_V

  • F specifies training framework: cf is Caffe, tf is Tensorflow, dk is Darknet, pt is PyTorch
  • M specifies the model
  • D specifies the dataset
  • H specifies the height of input data
  • W specifies the width of input data
  • P specifies the pruning ratio, it means how much computation is reduced. It is optional depending on whether the model is pruned.
  • C specifies the computation of the model: how many Gops per image
  • V specifies the version of Vitis AI

For example, cf_refinedet_coco_480_360_0.8_25G_1.1 is a RefineDet model trained with Caffe using COCO dataset, input data size is 480*360, 80% pruned, the computation per image is 25Gops and Vitis AI version is 1.1.

caffe-xilinx

This is a custom distribution of caffe. Please use caffe-xilinx to test/finetune the caffe models listed in this page.

Note: To download caffe-xlinx, visit caffe-xilinx.zip

Model Download

The following table lists various models, download link and MD5 checksum for the zip file of each model.

Note: To download all the models, visit all_models_1.1.zip.

Click here to view details

If you are a:

  • Linux user, use the get_model.sh script to download all the models.
  • Windows user, use the download link listed in the following table to download a model.
No. Model Size Download link Checksum
1 cf_resnet50_imagenet_224_224_7.7G_1.1 203.99 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_resnet50_imagenet_224_224_1.1.zip fe1fcbbdc935dc5cdf75a95780f8983e
2 cf_inceptionv1_imagenet_224_224_3.16G_1.1 79.41 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_inceptionv1_imagenet_224_224_1.1.zip 17417893abee5c489c25de0420d927b6
3 cf_inceptionv2_imagenet_224_224_4G_1.1 128.63 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_inceptionv2_imagenet_224_224_1.1.zip ae52f235af9f1e21aee8aa20d68905f5
4 cf_inceptionv3_imagenet_299_299_11.4G_1.1 190.74 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_inceptionv3_imagenet_299_299_1.1.zip b44315d620c4a9e5266763f34f44b7c8
5 cf_inceptionv4_imagenet_299_299_24.5G_1.1 341.64 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_inceptionv4_imagenet_299_299_1.1.zip 5df3ae9c4daf6f3276612de579239359
6 cf_mobilenetv2_imagenet_224_224_0.59G_1.1 19.67 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_mobilenetv2_imagenet_224_224_1.1.zip 1dbae9a4a8f968ffba665fe67af1ceb6
7 cf_squeeze_imagenet_227_227_0.76G_1.1 10.04 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_squeeze_imagenet_227_227_1.1.zip 9da53315ea5bfdde385c57e3a451afd5
8 cf_resnet18_imagenet_224_224_3.65G_1.1 130.7MB https://www.xilinx.com/bin/public/openDownload?filename=cf_resnet18_imagenet_224_224_1.1.zip f8a926550af500b0848db74e8e2e1381
9 cf_ssdpedestrian_coco_360_640_0.97_5.9G_1.1 5.96 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_ssdpedestrian_coco_360_640_0.97_1.1.zip 0f9ad09d11e250b8a43aa110084090a2
10 cf_refinedet_coco_360_480_0.8_25G_1.1 92.9 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_refinedet_coco_480_360_0.8_1.1.zip a0a5c91b9f641c71727b786ea22ae018
11 cf_refinedet_coco_360_480_0.92_10.10G_1.1 8.12 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_refinedet_coco_480_360_0.92_1.1.zip 02cbc7ea6b8c2b668af3e6f912303315
12 cf_refinedet_coco_360_480_0.96_5.08G_1.1 4.28 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_refinedet_coco_480_360_0.96_1.1.zip 1756e7e8c9ec30e5948076acadbed811
13 cf_ssdadas_bdd_360_480_0.95_6.3G_1.1 10.71 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_ssdadas_bdd_360_480_0.95_1.1.zip e5a0f9b3b6e2c72aa8961e279a3cef11
14 cf_ssdtraffic_360_480_0.9_11.6G_1.1 15.62 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_ssdtraffic_360_480_0.9_1.1.zip a8d6a9db2bb40b16cc1de435709bf570
15 cf_VPGnet_caltechlane_480_640_0.99_2.5G_1.1 5.97 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_VPGnet_caltechlane_480_640_0.99_1.1.zip 02bd13eaa6d7d4e1b5a4fd0280c8d2e1
16 cf_ssdmobilenetv2_bdd_360_480_6.57G_1.1 77 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_ssdmobilenetv2_bdd_360_480_1.1.zip 868b430065d39cf60f3ca0662e8e8b9e
17 cf_fpn_cityscapes_256_512_8.9G_1.1 69.27MB https://www.xilinx.com/bin/public/openDownload?filename=cf_fpn_cityscapes_256_512_1.1.zip 37b49ee32d0b8974ce2787fff6eebca2
18 cf_SPnet_aichallenger_224_128_0.54G_1.1 11 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_SPnet_aichallenger_224_128_1.1.zip 13f4c7702b94aeb7bcb2dfecc29c5e87
19 cf_openpose_aichallenger_368_368_0.3_189.7G_1.1 408.42 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_openpose_aichallenger_368_368_0.3_1.1.zip 446b135dc57408868f49de7a61142dd7
20 cf_densebox_wider_320_320_0.49G_1.1 4.33 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_densebox_wider_320_320_1.1.zip 83ca356836a21d37affd97b129b636ff
21 cf_densebox_wider_360_640_1.11G_1.1 4.33 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_densebox_wider_360_640_1.1.zip 7f2e1e2599260a5dd0ab43df6ee1ca61
22 cf_landmark_celeba_96_72_0.14G_1.1 45.25 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_landmark_celeba_96_72_1.1.zip 8adfd25d1a3225fe0fdd2152c170e4e3
23 cf_reid_market1501_160_80_0.95G_1.1 88.96 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_reid_market1501_160_80_1.1.zip 4ef83b5e9ef601d3ad99ca71b19bab26
24 cf_multitask_bdd_288_512_14.8G_1.1 108.75 MB https://www.xilinx.com/bin/public/openDownload?filename=cf_multitask_bdd_288_512_1.1.zip 3d35e68e71cd8911e40c3d57fb7328e5
25 dk_yolov3_bdd_288_512_53.7G_1.1 695.59 MB https://www.xilinx.com/bin/public/openDownload?filename=dk_yolov3_bdd_288_512_1.1.zip 346a0810ed3f5d8f004f74934d9a8464
26 dk_yolov3_cityscapes_256_512_0.9_5.46G_1.1 26.96 MB https://www.xilinx.com/bin/public/openDownload?filename=dk_yolov3_cityscapes_256_512_0.9_1.1.zip 63e1a349a35188e9e08136072750690e
27 dk_yolov3_voc_416_416_65.42G_1.1 689.58 MB https://www.xilinx.com/bin/public/openDownload?filename=dk_yolov3_voc_416_416_1.1.zip 7bb0afd4766f46d298e07fcc595a6f4a
28 dk_yolov2_voc_448_448_34G_1.1 410.57 MB https://www.xilinx.com/bin/public/openDownload?filename=dk_yolov2_voc_448_448_1.1.zip 8f7c48f9c5da0d652fef7091c1c11c34
29 dk_yolov2_voc_448_448_0.66_11.56G_1.1 167.56 MB https://www.xilinx.com/bin/public/openDownload?filename=dk_yolov2_voc_448_448_0.66_1.1.zip 2184ea9b85c94dd04cdd05c4556bab8e
30 dk_yolov2_voc_448_448_0.71_9.86G_1.1 151.83 MB https://www.xilinx.com/bin/public/openDownload?filename=dk_yolov2_voc_448_448_0.71_1.1.zip 33b830f11b6f89d351b465705cd67e3f
31 dk_yolov2_voc_448_448_0.77_7.82G_1.1 110.02 MB https://www.xilinx.com/bin/public/openDownload?filename=dk_yolov2_voc_448_448_0.77_1.1.zip 1619814a35f439836ef69a5c34249af1
32 tf_inceptionresnetv2_imagenet_299_299_26.35G_1.1 446.69 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_inception_resnet_v2_imagenet_299_299_1.1.zip 95b6f5bbc7c0b7772fc55963e2d4bb47
33 tf_inceptionv1_imagenet_224_224_3G_1.1 53.43 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_inceptionv1_imagenet_224_224_1.1.zip 935705a7a371f508953128ba3d07fda5
34 tf_inceptionv3_imagenet_299_299_11.45G_1.1 191.14 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_inceptionv3_imagenet_299_299_1.1.zip c12517b59ab5d05c40f501072bfbd270
35 tf_inceptionv4_imagenet_299_299_24.55G_1.1 342.44 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_inceptionv4_imagenet_299_299_1.1.zip b8dc266e3261c32ebc18de0be9a84a00
36 tf_mobilenetv1_0.25_imagenet_128_128_27.15M_1.1 3.84 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_mobilenetv1_0.25_imagenet_128_128_1.1.zip 90c3f9e0fe33c52f4bd5e3c33e78c22d
37 tf_mobilenetv1_0.5_imagenet_160_160_150.07M_1.1 10.67 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_mobilenetv1_0.5_imagenet_160_160_1.1.zip 132e6ac454918ae82e0024ee976a2c49
38 tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1 33.38 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_mobilenetv1_1.0_imagenet_224_224_1.1.zip 077deb4d3fdf2d42765b8fe55e251f76
39 tf_mobilenetv2_1.0_imagenet_224_224_0.59G_1.1 28.31 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_mobilenetv2_1.0_imagenet_224_224_1.1.zip 032ca92d995ff4b8dc654566f48024b0
40 tf_mobilenetv2_1.4_imagenet_224_224_1.16G_1.1 49.07 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_mobilenetv2_1.4_imagenet_224_224_1.1.zip 80d41f50d67d4674d63852968675d462
41 tf_resnetv1_50_imagenet_224_224_6.97G_1.1 204.47 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_resnetv1_50_imagenet_224_224_1.1.zip 56336c2b37584bac07b0c5aa2b6e408e
42 tf_resnetv1_101_imagenet_224_224_14.4G_1.1 356.08 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_resnetv1_101_imagenet_224_224_1.1.zip 3fccc5a329fd77bf82717454628e78e4
43 tf_resnetv1_152_imagenet_224_224_21.83G_1.1 481.89 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_resnetv1_152_imagenet_224_224_1.1.zip 8350c88145eac9bc82a7f5a1998580a3
44 tf_vgg16_imagenet_224_224_30.96G_1.1 1.1 GB https://www.xilinx.com/bin/public/openDownload?filename=tf_vgg16_imagenet_224_224_1.1.zip be883c1bf5be7e5ee889144035e48600
45 tf_vgg19_imagenet_224_224_39.28G_1.1 1.14 GB https://www.xilinx.com/bin/public/openDownload?filename=tf_vgg19_imagenet_224_224_1.1.zip 885ec350b6d913e5ddd3f1b38190440b
46 tf_ssdmobilenetv1_coco_300_300_2.47G_1.1 53.59 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_ssdmobilenetv1_coco_300_300_1.1.zip 7693181b6a74bfc3eb6edf93d2e484ff
47 tf_ssdmobilenetv2_coco_300_300_3.75G_1.1 129.96 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_ssdmobilenetv2_coco_300_300_1.1.zip 1f2efc1eb61d4bf43ca814944fd20c6f
48 tf_ssdresnet50v1_fpn_coco_640_640_178.4G_1.1 373.07 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_ssdresnet50_fpn_coco_640_640_1.1.zip f9b4832e9abe06c8dffff2b6d6e12323
49 tf_yolov3_voc_416_416_65.63G_1.1 500.1 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_yolov3_voc_416_416_1.1.zip 3c592d349dfeb0c8807409e34cce9145
50 tf_mlperf_resnet34_coco_1200_1200_433G_1.1 236.3 MB https://www.xilinx.com/bin/public/openDownload?filename=tf_mlperf_resnet34_coco_1200_1200_1.1.zip 29a689286036842c34405b42372ad6a3
- all_models_1.1 9.86 GB https://www.xilinx.com/bin/public/openDownload?filename=all_models_1.1.zip 10bedfa99692c5d0e7f840d23d0cd8d0

Model Directory Structure

Download and extract the model archive to your working area on the local hard disk. For details on the various models, their download link and MD5 checksum for the zip file of each model, see Model Download.

caffe Model Directory Structure

For a caffe model, you should see the following directory structure:

├── code                            # Contains code
│   ├── test                        # Contains test code which can run demo and evaluate model performance.
│   └── train                       # Contains training code
│                                     
│                                   
├── readme.md                       # Contains the environment requirements, data preprocess and model information.
│                                     Refer this to know that how to test and train the model with scripts.
│                                        
├── data                            # Contains the dataset that used for model test and training.
│                                     When test or training script runs successfully, dataset will be automatically placed in it.
│                                                       
├── quantized                             
│   ├── deploy.caffemodel           # Quantized weights, the output of vai_q_caffe without modification.
│   ├── deploy.prototxt             # Quantized prototxt, the output of vai_q_caffe without modification.
│   ├── quantized_test.prototxt     # Used to run evaluation with quantized_train_test.caffemodel on GPU
│   │                                 using python test code released in near future. Some models don't have this file
│   │                                 if they are converted from Darknet (Yolov2, Yolov3),
│   │                                 Pytorch (ReID) or there is no Caffe Test (Densebox).                                 
│   ├── quantized_train_test.caffemodel   # Quantized weights can be used for quantizeded-point training and evaluation.    
│   └── quantized_train_test.prototxt     # Used for quantized-point training and testing with quantized_train_test.caffemodel
│                                           on GPU when datalayer modified to user's data path.
└── float                           
    ├── float.caffemodel            # Trained float-point weights.
    ├── float.prototxt              # Modified test.prototxt as the input to vai_q_caffe along
    │                                 with float.caffemodel. vai_q_caffe is Xilinx quantization tool
    │                                 which quantizes float-point to quantized-point model with minimal
    │                                 accuracy loss.
    ├── test.prototxt               # Used to run evaluation with python test codes released in near future.    
    └── trainval.prorotxt           # Used for training and testing with caffe train/test command
                                      when datalayer modified to user's data path. Some models don't
                                      have this file if they are converted from Darknet (Yolov2, Yolov3),
                                      Pytorch (ReID) or there is no Caffe Test (Densebox).

Tensorflow Model Directory Structure

For a Tensorflow model, you should see the following directory structure:

├── code                            # Contains code
│   └── test                        # Contains test code which can run demo and evaluate model performance.
│
├── readme.md                       # Contains the environment requirements, data preprocess and model information.
│                                     Refer this to know that how to test the model with scripts.
│
├── data                            # Contains the dataset that used for model test and training.
│                                     When test or training script runs successfully, dataset will be automatically placed in it.
│
├── quantized                          
│   ├── deploy.model.pb             # Quantized model for the compiler (extended Tensorflow format).
│   └── quantize_eval_model.pb      # Quantized model for evaluation.
│
└── float                             
    └── frozen.pb                   # Float-point frozen model, the input to the `vai_q_tensorflow`.

Note: For more information on vai_q_caffe and vai_q_tensorflow, see the Vitis AI User Guide.

Model Performance

All the models in the Model Zoo have been deployed on Xilinx hardware with Vitis AI and Vitis AI Library. The performance number including end-to-end throughput and latency for each model on various boards with different DPU configurations are listed in the following sections.

For more information about DPU, see DPU IP Product Guide.

Note: The model performance number listed in the following sections is generated with Vitis AI v1.1 and Vitis AI Lirary v1.1. For each board, a different DPU configuration is used. Vitis AI and Vitis AI Library can be downloaded for free from Vitis AI Github and Vitis AI Library Github. We will continue to improve the performance with Vitis AI. The performance number reported here is subject to change in the near future.

Performance on ZCU102 (0432055-04)

This version of ZCU102 is out of stock. The performance number shown below was measured with the previous AI SDK v2.0.4.

Click here to view details

The following table lists the performance number including end-to-end throughput and latency for each model on the ZCU102 (0432055-04) board with a 3 * B4096 @ 287MHz V1.4.0 DPU configuration:

No. Model Name E2E latency (ms) Thread num =1 E2E throughput -fps(Single Thread) E2E throughput -fps(Multi Thread)
1 resnet50 cf_resnet50_imagenet_224_224_7.7G 12.85 77.8 179.3
2 Inception_v1 cf_inceptionv1_imagenet_224_224_3.16G 5.47 182.683 485.533
3 Inception_v2 cf_inceptionv2_imagenet_224_224_4G 6.76 147.933 373.267
4 Inception_v3 cf_inceptionv3_imagenet_299_299_11.4G 17 58.8333 155.4
5 mobileNet_v2 cf_mobilenetv2_imagenet_224_224_0.59G 4.09 244.617 638.067
6 tf_resnet50 tf_resnet50_imagenet_224_224_6.97G 11.94 83.7833 191.417
7 tf_inception_v1 tf_inceptionv1_imagenet_224_224_3G 6.72 148.867 358.283
8 tf_mobilenet_v2 tf_mobilenetv2_imagenet_224_224_1.17G 5.46 183.117 458.65
9 ssd_adas_pruned_0.95 cf_ssdadas_bdd_360_480_0.95_6.3G 11.33 88.2667 320.5
10 ssd_pedestrain_pruned_0.97 cf_ssdpedestrian_coco_360_640_0.97_5.9G 12.96 77.1833 314.717
11 ssd_traffic_pruned_0.9 cf_ssdtraffic_360_480_0.9_11.6G 17.49 57.1833 218.183
12 ssd_mobilnet_v2 cf_ssdmobilenetv2_bdd_360_480_6.57G 24.21 41.3 141.233
13 tf_ssd_voc tf_ssd_voc_300_300_64.81G 69.28 14.4333 46.7833
14 densebox_320_320 cf_densebox_wider_320_320_0.49G 2.43 412.183 1416.63
15 densebox_360_640 cf_densebox_wider_360_640_1.11G 5.01 199.717 719.75
16 yolov3_adas_prune_0.9 dk_yolov3_cityscapes_256_512_0.9_5.46G 11.09 90.1667 259.65
17 yolov3_voc dk_yolov3_voc_416_416_65.42G 70.51 14.1833 44.4
18 tf_yolov3_voc tf_yolov3_voc_416_416_65.63G 70.75 14.1333 44.0167
19 refinedet_pruned_0.8 cf_refinedet_coco_360_480_0.8_25G 29.91 33.4333 109.067
20 refinedet_pruned_0.92 cf_refinedet_coco_360_480_0.92_10.10G 15.39 64.9667 216.317
21 refinedet_pruned_0.96 cf_refinedet_coco_360_480_0.96_5.08G 11.04 90.5833 312
22 FPN cf_fpn_cityscapes_256_512_8.9G 16.58 60.3 203.867
23 VPGnet_pruned_0.99 cf_VPGnet_caltechlane_480_640_0.99_2.5G 9.44 105.9 424.667
24 SP-net cf_SPnet_aichallenger_224_128_0.54G 1.73 579.067 1620.67
25 Openpose_pruned_0.3 cf_openpose_aichallenger_368_368_0.3_189.7G 279.07 3.58333 38.5
26 yolov2_voc dk_yolov2_voc_448_448_34G 39.76 25.15 86.35
27 yolov2_voc_pruned_0.66 dk_yolov2_voc_448_448_0.66_11.56G 18.42 54.2833 211.217
28 yolov2_voc_pruned_0.71 dk_yolov2_voc_448_448_0.71_9.86G 16.42 60.9167 242.433
29 yolov2_voc_pruned_0.77 dk_yolov2_voc_448_448_0.77_7.82G 14.46 69.1667 286.733
30 Inception-v4 cf_inceptionv4_imagenet_299_299_24.5G 34.25 29.2 84.25
31 SqueezeNet cf_squeeze_imagenet_227_227_0.76G 3.6 277.65 1080.77
32 face_landmark cf_landmark_celeba_96_72_0.14G 1.13 885.033 1623.3
33 reid cf_reid_marketcuhk_160_80_0.95G 2.67 375 773.533
34 yolov3_bdd dk_yolov3_bdd_288_512_53.7G 73.89 13.5333 42.8833
35 tf_mobilenet_v1 tf_mobilenetv1_imagenet_224_224_1.14G 3.2 312.067 875.967
36 resnet18 cf_resnet18_imagenet_224_224_3.65G 5.1 195.95 524.433
37 resnet18_wide tf_resnet18_imagenet_224_224_28G 33.28 30.05 83.4167

Performance on ZCU102 (0432055-05)

Measured with Vitis AI 1.1 and Vitis AI Library 1.1

Click here to view details

The following table lists the performance number including end-to-end throughput and latency for each model on the ZCU102 (0432055-05) board with a 3 * B4096 @ 281MHz V1.4.1 DPU configuration:

No. Model Name E2E latency (ms) Thread num =1 E2E throughput -fps(Single Thread) E2E throughput -fps(Multi Thread)
1 resnet50 cf_resnet50_imagenet_224_224_7.7G 13.74 72.8 155.5
2 resnet18 cf_resnet18_imagenet_224_224_3.65G 5.72 174.7 461.6
3 Inception_v1 cf_inceptionv1_imagenet_224_224_3.16G 6.00 166.6 444.0
4 Inception_v2 cf_inceptionv2_imagenet_224_224_4G 7.31 136.7 335.6
5 Inception_v3 cf_inceptionv3_imagenet_299_299_11.4G 17.84 56.0 138.5
6 Inception_v4 cf_inceptionv4_imagenet_299_299_24.5G 35.54 28.1 71.3
7 Mobilenet_v2 cf_mobilenetv2_imagenet_224_224_0.59G 4.66 214.7 580.6
8 SqueezeNet cf_squeeze_imagenet_227_227_0.76G 3.71 269.4 1045.9
9 ssd_pedestrain_pruned_0_97 cf_ssdpedestrian_coco_360_640_0.97_5.9G 13.19 75.7 294.4
10 refinedet_pruned_0_8 cf_refinedet_coco_360_480_0.8_25G 31.57 31.7 103.8
11 refinedet_pruned_0_92 cf_refinedet_coco_360_480_0.92_10.10G 16.71 59.8 204.0
12 refinedet_pruned_0_96 cf_refinedet_coco_360_480_0.96_5.08G 12.10 82.6 290.1
13 ssd_adas_pruned_0_95 cf_ssdadas_bdd_360_480_0.95_6.3G 12.10 82.6 296.3
14 ssd_traffic_pruned_0_9 cf_ssdtraffic_360_480_0.9_11.6G 18.39 54.4 206.7
15 VPGnet_pruned_0_99 cf_VPGnet_caltechlane_480_640_0.99_2.5G 9.77 102.3 397.3
16 ssd_mobilenet_v2 cf_ssdmobilenetv2_bdd_360_480_6.57G 26.04 38.4 112.5
17 FPN cf_fpn_cityscapes_256_512_8.9G 16.74 59.7 185.2
18 SP_net cf_SPnet_aichallenger_224_128_0.54G 2.03 491.6 1422.8
19 Openpose_pruned_0_3 cf_openpose_aichallenger_368_368_0.3_189.7G 286.10 3.5 15.3
20 densebox_320_320 cf_densebox_wider_320_320_0.49G 2.57 388.3 1279
21 densebox_640_360 cf_densebox_wider_360_640_1.11G 5.13 195.0 627.8
22 face_landmark cf_landmark_celeba_96_72_0.14G 1.18 846.7 1379.9
23 reid cf_reid_market1501_160_80_0.95G 2.76 361.9 672.8
24 multi_task cf_multitask_bdd_288_512_14.8G 28.26 35.4 133.0
25 yolov3_bdd dk_yolov3_bdd_288_512_53.7G 77.12 13.0 37.1
26 yolov3_adas_pruned_0_9 dk_yolov3_cityscapes_256_512_0.9_5.46G 11.93 83.8 235.3
27 yolov3_voc dk_yolov3_voc_416_416_65.42G 73.82 13.5 38.2
28 yolov2_voc dk_yolov2_voc_448_448_34G 40.28 24.8 77.1
29 yolov2_voc_pruned_0_66 dk_yolov2_voc_448_448_0.66_11.56G 18.78 53.2 194.2
30 yolov2_voc_pruned_0_71 dk_yolov2_voc_448_448_0.71_9.86G 16.71 59.8 224.0
31 yolov2_voc_pruned_0_77 dk_yolov2_voc_448_448_0.77_7.82G 14.71 68.0 266.2
32 Inception_resnet_v2 tf_inceptionresnetv2_imagenet_299_299_26.35G 42.81 23.3 51.0
33 Inception_v1 tf_inceptionv1_imagenet_224_224_3G 5.81 172.1 455.6
34 Inception_v3 tf_inceptionv3_imagenet_299_299_11.45G 17.90 55.9 136.8
35 Inception_v4 tf_inceptionv4_imagenet_299_299_24.55G 35.56 28.1 71.4
36 Mobilenet_v1 tf_mobilenetv1_0.25_imagenet_128_128_27.15M 1.18 848.6 2260.9
37 Mobilenet_v1 tf_mobilenetv1_0.5_imagenet_160_160_150.07M 1.73 577.2 1913.7
38 Mobilenet_v1 tf_mobilenetv1_1.0_imagenet_224_224_1.14G 3.83 261.3 788.6
39 Mobilenet_v2 tf_mobilenetv2_1.0_imagenet_224_224_0.59G 4.57 218.6 598.7
40 Mobilenet_v2 tf_mobilenetv2_1.4_imagenet_224_224_1.16G 6.15 162.4 412.0
41 resnet_v1_50 tf_resnetv1_50_imagenet_224_224_6.97G 12.76 78.3 164.7
42 resnet_v1_101 tf_resnetv1_101_imagenet_224_224_14.4G 23.00 43.5 94.5
43 resnet_v1_152 tf_resnetv1_152_imagenet_224_224_21.83G 33.43 29.9 66.1
44 vgg_16 tf_vgg16_imagenet_224_224_30.96G 50.43 19.8 44.7
45 vgg_19 tf_vgg19_imagenet_224_224_39.28G 58.35 17.1 40.3
46 ssd_mobilenet_v1 tf_ssdmobilenetv1_coco_300_300_2.47G 11.46 87.2 323.5
47 ssd_mobilenet_v2 tf_ssdmobilenetv2_coco_300_300_3.75G 15.85 63.0 198.7
48 ssd_resnet_50_v1_fpn tf_ssdresnet50v1_fpn_coco_640_640_178.4G 745.55 1.3 5.0
49 yolov3_voc tf_yolov3_voc_416_416_65.63G 74.24 13.5 37.8
50 mlperf_ssd_resnet34 tf_mlperf_resnet34_coco_1200_1200_433G 547.28 1.8 7.4

Performance on ZCU104

Measured with Vitis AI 1.1 and Vitis AI Library 1.1

Click here to view details

The following table lists the performance number including end-to-end throughput and latency for each model on the ZCU104 board with a 2 * B4096 @ 300MHz V1.4.1 DPU configuration:

No. Model Name E2E latency (ms) Thread num =1 E2E throughput -fps(Single Thread) E2E throughput -fps(Multi Thread)
1 resnet50 cf_resnet50_imagenet_224_224_7.7G 12.46 80.2 146.8
2 resnet18 cf_resnet18_imagenet_224_224_3.65G 5.08 196.7 403.7
3 Inception_v1 cf_inceptionv1_imagenet_224_224_3.16G 5.27 189.8 387.0
4 Inception_v2 cf_inceptionv2_imagenet_224_224_4G 6.55 152.7 298.2
5 Inception_v3 cf_inceptionv3_imagenet_299_299_11.4G 16.51 60.5 117.3
6 Inception_v4 cf_inceptionv4_imagenet_299_299_24.5G 33.10 30.2 58.6
7 Mobilenet_v2 cf_mobilenetv2_imagenet_224_224_0.59G 4.01 249.3 536.6
8 SqueezeNet cf_squeeze_imagenet_227_227_0.76G 3.64 274.4 941.8
9 ssd_pedestrain_pruned_0_97 cf_ssdpedestrian_coco_360_640_0.97_5.9G 12.80 78.1 221.5
10 refinedet_pruned_0_8 cf_refinedet_coco_360_480_0.8_25G 30.68 32.6 76.1
11 refinedet_pruned_0_92 cf_refinedet_coco_360_480_0.92_10.10G 16.32 61.3 154.2
12 refinedet_pruned_0_96 cf_refinedet_coco_360_480_0.96_5.08G 11.96 83.6 228.7
13 ssd_adas_pruned_0_95 cf_ssdadas_bdd_360_480_0.95_6.3G 11.80 84.7 231.9
14 ssd_traffic_pruned_0_9 cf_ssdtraffic_360_480_0.9_11.6G 17.80 56.1 153.2
15 VPGnet_pruned_0_99 cf_VPGnet_caltechlane_480_640_0.99_2.5G 9.47 105.5 364.9
16 ssd_mobilenet_v2 cf_ssdmobilenetv2_bdd_360_480_6.57G 39.29 25.4 101.3
17 FPN cf_fpn_cityscapes_256_512_8.9G 16.12 62 169.9
18 SP_net cf_SPnet_aichallenger_224_128_0.54G 1.81 552.4 1245.6
19 Openpose_pruned_0_3 cf_openpose_aichallenger_368_368_0.3_189.7G 274.46 3.6 11.0
20 densebox_320_320 cf_densebox_wider_320_320_0.49G 2.52 397.4 1250.3
21 densebox_640_360 cf_densebox_wider_360_640_1.11G 5.03 198.7 606.6
22 face_landmark cf_landmark_celeba_96_72_0.14G 1.12 890.1 1363.2
23 reid cf_reid_market1501_160_80_0.95G 2.59 385.6 668.8
24 multi_task cf_multitask_bdd_288_512_14.8G 27.76 36.0 108.4
25 yolov3_bdd dk_yolov3_bdd_288_512_53.7G 73.62 13.6 28.7
26 yolov3_adas_pruned_0_9 dk_yolov3_cityscapes_256_512_0.9_5.46G 11.83 84.5 218.5
27 yolov3_voc dk_yolov3_voc_416_416_65.42G 70.27 14.2 29.5
28 yolov2_voc dk_yolov2_voc_448_448_34G 38.17 26.2 59.1
29 yolov2_voc_pruned_0_66 dk_yolov2_voc_448_448_0.66_11.56G 17.99 55.6 153.2
30 yolov2_voc_pruned_0_71 dk_yolov2_voc_448_448_0.71_9.86G 16.02 62.4 180.2
31 yolov2_voc_pruned_0_77 dk_yolov2_voc_448_448_0.77_7.82G 14.17 70.5 217.4
32 Inception_resnet_v2 tf_inceptionresnetv2_imagenet_299_299_26.35G 39.35 25.4 46.1
33 Inception_v1 tf_inceptionv1_imagenet_224_224_3G 5.10 196.1 401.7
34 Inception_v3 tf_inceptionv3_imagenet_299_299_11.45G 16.57 60.3 116.4
35 Inception_v4 tf_inceptionv4_imagenet_299_299_24.55G 33.13 30.2 58.6
36 Mobilenet_v1 tf_mobilenetv1_0.25_imagenet_128_128_27.15M 0.79 1263.6 3957.7
37 Mobilenet_v1 tf_mobilenetv1_0.5_imagenet_160_160_150.07M 1.31 763.1 2038.1
38 Mobilenet_v1 tf_mobilenetv1_1.0_imagenet_224_224_1.14G 3.21 311.8 731.1
39 Mobilenet_v2 tf_mobilenetv2_1.0_imagenet_224_224_0.59G 3.98 250.9 546.6
40 Mobilenet_v2 tf_mobilenetv2_1.4_imagenet_224_224_1.16G 5.39 185.5 381.5
41 resnet_v1_50 tf_resnetv1_50_imagenet_224_224_6.97G 11.59 86.3 157.5
42 resnet_v1_101 tf_resnetv1_101_imagenet_224_224_14.4G 21.15 47.3 87.2
43 resnet_v1_152 tf_resnetv1_152_imagenet_224_224_21.83G 30.80 32.5 60.1
44 vgg_16 tf_vgg16_imagenet_224_224_30.96G 46.99 21.3 38.3
45 vgg_19 tf_vgg19_imagenet_224_224_39.28G 54.41 18.4 33.8
46 ssd_mobilenet_v1 tf_ssdmobilenetv1_coco_300_300_2.47G 10.82 92.4 330.0
47 ssd_mobilenet_v2 tf_ssdmobilenetv2_coco_300_300_3.75G 14.99 66.7 185.0
48 ssd_resnet50_v1_fpn tf_ssdresnet50v1_fpn_coco_640_640_178.4G 747.43 1.3 5.1
49 yolov3_voc tf_yolov3_voc_416_416_65.63G 70.64 14.1 29.3
50 mlperf_ssd_resnet34 tf_mlperf_resnet34_coco_1200_1200_433G 626.09 1.6 5.3

Performance on U50

Measured with Vitis AI 1.1 and Vitis AI Library 1.1

Click here to view details

The following table lists the performance number including end-to-end throughput and latency for each model on the Alveo U50 board with 6 DPUv3E kernels running at 250Mhz in Gen3x16:

No. Model Name E2E latency (ms) Thread num =1 E2E throughput -fps(Single Thread) E2E throughput -fps(Multi Thread)
1 resnet50 cf_resnet50_imagenet_224_224_7.7G 18.00 166.4 394
2 resnet18 cf_resnet18_imagenet_224_224_3.65G 8.95 334.6 995.1
3 Inception_v1 cf_inceptionv1_imagenet_224_224_3.16G 14.13 212.1 551
4 Inception_v2 cf_inceptionv2_imagenet_224_224_4G 17.07 175.5 426.4
5 Inception_v3 cf_inceptionv3_imagenet_299_299_11.4G 49.55 60.5 133.3
6 Inception_v4 cf_inceptionv4_imagenet_299_299_24.5G 101.98 29.4 61.5
7 SqueezeNet cf_squeeze_imagenet_227_227_0.76G 18.01 166.3 418
8 ssd_pedestrain_pruned_0_97 cf_ssdpedestrian_coco_360_640_0.97_5.9G 88.34 33.9 83.1
9 refinedet_pruned_0_8 cf_refinedet_coco_360_480_0.8_25G 104.57 28.7 64.4
10 refinedet_pruned_0_92 cf_refinedet_coco_360_480_0.92_10.10G 72.56 41.3 97.6
11 refinedet_pruned_0_96 cf_refinedet_coco_360_480_0.96_5.08G 73.04 41 96.8
12 ssd_adas_pruned_0_95 cf_ssdadas_bdd_360_480_0.95_6.3G 64.45 46.5 118.1
13 ssd_traffic_pruned_0_9 cf_ssdtraffic_360_480_0.9_11.6G 82.49 36.3 91.1
14 VPGnet_pruned_0_99 cf_VPGnet_caltechlane_480_640_0.99_2.5G 105.83 28.3 65.1
15 FPN cf_fpn_cityscapes_256_512_8.9G 80.24 37.3 116.9
16 SP_net cf_SPnet_aichallenger_224_128_0.54G 7.39 405.5 1074.5
17 Openpose_pruned_0_3 cf_openpose_aichallenger_368_368_0.3_189.7G 342.87 8.7 22.3
18 densebox_320_320 cf_densebox_wider_320_320_0.49G 12.57 238.4 796.2
19 densebox_640_360 cf_densebox_wider_360_640_1.11G 29.00 103.3 360.6
20 face_landmark cf_landmark_celeba_96_72_0.14G 1.42 2107.7 6631.9
21 reid cf_reid_market1501_160_80_0.95G 3.99 751.4 2301
22 multi_task cf_multitask_bdd_288_512_14.8G 120.52 24.9 70.8
23 yolov3_bdd dk_yolov3_bdd_288_512_53.7G 155.79 19.2 42
24 yolov3_adas_pruned_0_9 dk_yolov3_cityscapes_256_512_0.9_5.46G 90.55 33.1 75.4
25 yolov3_voc dk_yolov3_voc_416_416_65.42G 122.24 24.5 54.9
26 yolov2_voc dk_yolov2_voc_448_448_34G 58.87 50.9 141.6
27 yolov2_voc_pruned_0_66 dk_yolov2_voc_448_448_0.66_11.56G 47.24 63.4 187.5
28 yolov2_voc_pruned_0_71 dk_yolov2_voc_448_448_0.71_9.86G 45.20 66.3 203
29 yolov2_voc_pruned_0_77 dk_yolov2_voc_448_448_0.77_7.82G 42.27 70.9 227.9
30 Inception_resnet_v2 tf_inceptionresnetv2_imagenet_299_299_26.35G 136.51 21.9 46.5
31 Inception_v1 tf_inceptionv1_imagenet_224_224_3G 13.16 227.7 682
32 Inception_v3 tf_inceptionv3_imagenet_299_299_11.45G 52.42 57.2 133.3
33 Inception_v4 tf_inceptionv4_imagenet_299_299_24.55G 105.11 28.5 61.3
34 resnet_v1_50 tf_resnetv1_50_imagenet_224_224_6.97G 17.63 169.9 460.7
35 resnet_v1_101 tf_resnetv1_101_imagenet_224_224_14.4G 28.56 104.9 247
36 resnet_v1_152 tf_resnetv1_152_imagenet_224_224_21.83G 40.49 74 165.8
37 vgg_16 tf_vgg16_imagenet_224_224_30.96G 48.11 62.3 137.9
38 vgg_19 tf_vgg19_imagenet_224_224_39.28G 56.95 52.6 114.4
39 ssd_resnet50_v1_fpn tf_ssdresnet50v1_fpn_coco_640_640_178.4G 525.11 5.7 15.2
40 yolov3_voc tf_yolov3_voc_416_416_65.63G 121.91 24.6 54.7

Performance on U200

Measured with Vitis AI 1.1 and Vitis AI Library 1.1

Click here to view details

The following table lists the performance number including end-to-end throughput and latency for each model on the Alveo U200 board with 2 DPUv1 kernels running at 350Mhz with xilinx_u200_xdma_201830_2 shell:

No. Model Name E2E latency (ms) Thread num =1 E2E throughput -fps(Single Thread) E2E throughput -fps(Multi Thread)
1 resnet50 cf_resnet50_imagenet_224_224_7.7G 2.13 470.6 561.3
2 resnet18 cf_resnet18_imagenet_224_224_3.65G 2.08 481 1157.8
3 Inception_v1 cf_inceptionv1_imagenet_224_224_3.16G 2.39 418.5 1449.4
4 Inception_v2 cf_inceptionv2_imagenet_224_224_4G 2.11 475.1 1129.2
5 Inception_v3 cf_inceptionv3_imagenet_299_299_11.4G 15.67 63.8 371.6
6 Inception_v4 cf_inceptionv4_imagenet_299_299_24.5G 10.77 92.8 221.2
7 SqueezeNet cf_squeeze_imagenet_227_227_0.76G 10.99 91 1157.1
8 densebox_320_320 cf_densebox_wider_320_320_0.49G 8.69 115.1 667.9
9 yolov3_bdd dk_yolov3_bdd_288_512_53.7G 14.53 68.8 75.9
10 yolov3_voc dk_yolov3_voc_416_416_65.42G 19.90 50.3 82.1

Performance on U250

Measured with Vitis AI 1.1 and Vitis AI Library 1.1

Click here to view details

The following table lists the performance number including end-to-end throughput and latency for each model on the Alveo U250 board with 4 DPUv1 kernels running at 350Mhz with xilinx_u250_xdma_201830_1 shell:

No. Model Name E2E latency (ms) Thread num =1 E2E throughput -fps(Single Thread) E2E throughput -fps(Multi Thread)
1 resnet50 cf_resnet50_imagenet_224_224_7.7G 1.68 595.5 1223.95
2 resnet18 cf_resnet18_imagenet_224_224_3.65G 1.67 600.5 2422.5
3 Inception_v1 cf_inceptionv1_imagenet_224_224_3.16G 1.93 517.1 4059.8
4 Inception_v2 cf_inceptionv2_imagenet_224_224_4G 1.65 607.8 2327.1
5 Inception_v3 cf_inceptionv3_imagenet_299_299_11.4G 6.18 161.8 743.8
6 Inception_v4 cf_inceptionv4_imagenet_299_299_24.5G 5.77 173.4 452.4
7 SqueezeNet cf_squeeze_imagenet_227_227_0.76G 5.44 183.7 2314.3
8 densebox_320_320 cf_densebox_wider_320_320_0.49G 7.43 167.2 898.5
9 yolov3_bdd dk_yolov3_bdd_288_512_53.7G 14.27 70.1 146.7
10 yolov3_voc dk_yolov3_voc_416_416_65.42G 9.46 105.7 139.4

Performance on Ultra96

The performance number shown below was measured with the previous AI SDK v2.0.4 on Ultra96 v1. The Vitis platform of Ultra96 v2 has not been released yet. So the performance numbers are therefore not reported for this Model Zoo release.

Click here to view details

The following table lists the performance number including end-to-end throughput and latency for each model on the Ultra96 board with a 1 * B1600 @ 287MHz V1.4.0 DPU configuration:

Note: The original power supply of Ultra96 is not designed for high performance AI workload. The board may occasionally hang to run few models, When multi-thread is used. For such situations, NA is specified in the following table.

No. Model Name E2E latency (ms) Thread num =1 E2E throughput -fps(Single Thread) E2E throughput -fps(Multi Thread)
1 resnet50 cf_resnet50_imagenet_224_224_7.7G 30.8 32.4667 33.4667
2 Inception_v1 cf_inceptionv1_imagenet_224_224_3.16G 13.98 71.55 75.0667
3 Inception_v2 cf_inceptionv2_imagenet_224_224_4G 17.16 58.2667 61.2833
4 Inception_v3 cf_inceptionv3_imagenet_299_299_11.4G 44.05 22.7 23.4333
5 mobileNet_v2 cf_mobilenetv2_imagenet_224_224_0.59G 7.34 136.183 NA
6 tf_resnet50 tf_resnet50_imagenet_224_224_6.97G 28.02 35.6833 36.6
7 tf_inception_v1 tf_inceptionv1_imagenet_224_224_3G 16.96 58.9667 61.2833
8 tf_mobilenet_v2 tf_mobilenetv2_imagenet_224_224_1.17G 10.17 98.3 104.25
9 ssd_adas_pruned_0.95 cf_ssdadas_bdd_360_480_0.95_6.3G 24.3 41.15 46.2
10 ssd_pedestrain_pruned_0.97 cf_ssdpedestrian_coco_360_640_0.97_5.9G 23.29 42.9333 50.8
11 ssd_traffic_pruned_0.9 cf_ssdtraffic_360_480_0.9_11.6G 35.5 28.1667 31.8
12 ssd_mobilnet_v2 cf_ssdmobilenetv2_bdd_360_480_6.57G 60.79 16.45 27.8167
13 tf_ssd_voc tf_ssd_voc_300_300_64.81G 186.92 5.35 5.81667
14 densebox_320_320 cf_densebox_wider_320_320_0.49G 4.17 239.883 334.167
15 densebox_360_640 cf_densebox_wider_360_640_1.11G 8.55 117 167.2
16 yolov3_adas_prune_0.9 dk_yolov3_cityscapes_256_512_0.9_5.46G 22.79 43.8833 49.6833
17 yolov3_voc dk_yolov3_voc_416_416_65.42G 185.19 5.4 5.53
18 tf_yolov3_voc tf_yolov3_voc_416_416_65.63G 199.34 5.01667 5.1
19 refinedet_pruned_0.8 cf_refinedet_coco_360_480_0.8_25G 66.37 15.0667 NA
20 refinedet_pruned_0.92 cf_refinedet_coco_360_480_0.92_10.10G 32.17 31.0883 33.6667
21 refinedet_pruned_0.96 cf_refinedet_coco_360_480_0.96_5.08G 20.29 49.2833 55.25
22 FPN cf_fpn_cityscapes_256_512_8.9G 36.34 27.5167 NA
23 VPGnet_pruned_0.99 cf_VPGnet_caltechlane_480_640_0.99_2.5G 13.9 71.9333 NA
24 SP-net cf_SPnet_aichallenger_224_128_0.54G 3.82 261.55 277.4
25 Openpose_pruned_0.3 cf_openpose_aichallenger_368_368_0.3_189.7G 560.75 1.78333 NA
26 yolov2_voc dk_yolov2_voc_448_448_34G 118.11 8.46667 8.9
27 yolov2_voc_pruned_0.66 dk_yolov2_voc_448_448_0.66_11.56G 37.5 26.6667 30.65
28 yolov2_voc_pruned_0.71 dk_yolov2_voc_448_448_0.71_9.86G 30.99 32.2667 38.35
29 yolov2_voc_pruned_0.77 dk_yolov2_voc_448_448_0.77_7.82G 26.29 38.03333 46.8333
30 Inception-v4 cf_inceptionv4_imagenet_299_299_24.5G 88.76 11.2667 11.5333
31 SqueezeNet cf_squeeze_imagenet_227_227_0.76G 5.96 167.867 283.583
32 face_landmark cf_landmark_celeba_96_72_0.14G 2.95 339.183 347.633
33 reid cf_reid_market1501_160_80_0.95G 6.28 159.15 166.633
34 yolov3_bdd dk_yolov3_bdd_288_512_53.7G 193.55 5.16667 5.31667
35 tf_mobilenet_v1 tf_mobilenetv1_imagenet_224_224_1.14G 5.97 167.567 186.55
36 resnet18 cf_resnet18_imagenet_224_224_3.65G 13.47 74.2167 77.8167
37 resnet18_wide tf_resnet18_imagenet_224_224_28G 97.72 10.2333 10.3833

Contributing

We welcome community contributions. When contributing to this repository, first discuss the change you wish to make via:

You can also submit a pull request with details on how to improve the product. Prior to submitting your pull request, ensure that you can build the product and run all the demos with your patch. In case of a larger feature, provide a relevant demo.

License

Xilinx AI Model Zoo is licensed under Apache License Version 2.0. By contributing to the project, you agree to the license and copyright terms therein and release your contribution under these terms.

Copyright© 2019 Xilinx

About

No description, website, or topics provided.

Resources

Readme

License

Apache-2.0 license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

1 tags

Packages

No packages published

Languages

  • Shell 100.0%