Skip to content

Latest commit

 

History

History
366 lines (347 loc) · 16.1 KB

mxnet_models_checklist.md

File metadata and controls

366 lines (347 loc) · 16.1 KB

Model validation and performance analysis status for MXNet

Public models

Image classification on ImageNet

Model Availability in OMZ (2023.02.24) Availability in the validation table
alexnet + +
darknet53 + +
densenet121 + +
densenet161 + +
densenet169 + +
densenet201 + +
googlenet + +
hrnet_w18_c + +
hrnet_w18_small_v1_c + +
hrnet_w18_small_v2_c + +
hrnet_w30_c + +
hrnet_w32_c + +
hrnet_w40_c + +
hrnet_w44_c + +
hrnet_w48_c + +
hrnet_w64_c + +
inceptionv3 + +
mobilenet0.25 + +
mobilenet0.5 + +
mobilenet0.75 + +
mobilenet1.0 + +
mobilenet1.0_int8 + +
mobilenetv2_0.25 + +
mobilenetv2_0.5 + +
mobilenetv2_0.75 + +
mobilenetv2_1.0 + +
mobilenetv3_large + +
mobilenetv3_small + +
residualattentionnet128 + Error (Parameter 'residualattentionmodel0_hybridsequential0_conv0_weight' has not been initialized)
residualattentionnet164 + Error (Parameter 'residualattentionmodel0_hybridsequential0_conv0_weight' has not been initialized)
residualattentionnet200 + Error (Parameter 'residualattentionmodel0_hybridsequential0_conv0_weight' has not been initialized)
residualattentionnet236 + Error (Parameter 'residualattentionmodel0_hybridsequential0_conv0_weight' has not been initialized)
residualattentionnet452 + Error (Parameter 'residualattentionmodel0_hybridsequential0_conv0_weight' has not been initialized)
residualattentionnet56 + Error (Parameter 'residualattentionmodel0_hybridsequential0_conv0_weight' has not been initialized)
residualattentionnet92 + Error (Parameter 'residualattentionmodel0_hybridsequential0_conv0_weight' has not been initialized)
resnest101 + +
resnest14 + +
resnest200 + +
resnest26 + +
resnest269 + +
resnest50 + +
resnet101_v1 + +
resnet101_v1b + +
resnet101_v1c + +
resnet101_v1d + +
resnet101_v1d_0.73 + +
resnet101_v1d_0.76 + +
resnet101_v1e + Error (Failed loading Parameter 'resnetv1e_batchnorm2_gamma' from saved params: shape incompatible expected (128,) vs saved (64,))
resnet101_v1s + +
resnet101_v2 + +
resnet152_v1 + +
resnet152_v1b + +
resnet152_v1c + +
resnet152_v1d + +
resnet152_v1e + Error (Failed loading Parameter 'resnetv1e_batchnorm2_gamma' from saved params: shape incompatible expected (128,) vs saved (64,))
resnet152_v1s + +
resnet152_v2 + +
resnet18_v1 + +
resnet18_v1b + +
resnet18_v1b_0.89 + +
resnet18_v1b_custom + +
resnet18_v2 + +
resnet34_v1 + +
resnet34_v1b + +
resnet34_v2 + +
resnet50_v1 + +
resnet50_v1_int8 + +
resnet50_v1b + +
resnet50_v1b_custom + +
resnet50_v1b_gn + +
resnet50_v1c + +
resnet50_v1d + +
resnet50_v1d_0.11 + +
resnet50_v1d_0.37 + +
resnet50_v1d_0.48 + +
resnet50_v1d_0.86 + +
resnet50_v1e + Error (Failed loading Parameter 'resnetv1e_batchnorm0_gamma' from saved params: shape incompatible expected (64,) vs saved (32,))
resnet50_v1s + +
resnet50_v2 + +
resnext101_32x4d + +
resnext101_64x4d + +
resnext50_32x4d + +
se_resnext101_32x4d + +
se_resnext101_64x4d + +
se_resnext50_32x4d + +
senet_154 + +
senet_154e + Error (Parameter 'features.11.0.downsample.1.weight' is missing...)
squeezenet1.0 + +
squeezenet1.1 + +
shufflenet_v1 + Error (Parameter 'shufflenetv10_conv0_weight' has not been initialized)
shufflenet_v2 + Error (Parameter 'shufflenetv20_conv0_weight' has not been initialized)
vgg11 + +
vgg11_bn + +
vgg13 + +
vgg13_bn + +
vgg16 + +
vgg16_bn + +
vgg19 + +
vgg19_bn + +
xception + +

Object detection

Model Availability in OMZ (2023.02.24) Availability in the validation table
custom_faster_rcnn_fpn + TypeError: custom_faster_rcnn_fpn() missing 1 required positional argument: 'classes'
center_net_dla34_coco + Pretrained model for center_net_dla34_coco is not available.
center_net_dla34_dcnv2_coco + Pretrained model for center_net_dla34_dcnv2_coco is not available.
center_net_dla34_dcnv2_voc + Pretrained model for center_net_dla34_dcnv2_voc is not available.
center_net_dla34_voc + Pretrained model for center_net_dla34_voc is not available.
center_net_mobilenetv3_large_duc_coco + +
center_net_mobilenetv3_large_duc_voc + +
center_net_mobilenetv3_small_duc_coco + +
center_net_mobilenetv3_small_duc_voc + +
center_net_resnet101_v1b_coco + +
center_net_resnet101_v1b_dcnv2_coco + +
center_net_resnet101_v1b_dcnv2_voc + +
center_net_resnet101_v1b_voc + +
center_net_resnet18_v1b_coco + +
center_net_resnet18_v1b_dcnv2_coco + +
center_net_resnet18_v1b_dcnv2_voc + +
center_net_resnet18_v1b_voc + +
center_net_resnet50_v1b_coco + +
center_net_resnet50_v1b_dcnv2_coco + +
center_net_resnet50_v1b_dcnv2_voc + +
center_net_resnet50_v1b_voc + +
doublehead_rcnn_resnet50_v1b_voc + RuntimeError: Parameter 'doubleheadrcnn0_double_fc_dense0_weight' has not been initialized.
dla34 + -
faster_rcnn_fpn_resnet101_v1d_coco + +
faster_rcnn_fpn_resnet50_v1b_coco + +
faster_rcnn_fpn_syncbn_resnest101_coco + +
faster_rcnn_fpn_syncbn_resnest269_coco + +
faster_rcnn_fpn_syncbn_resnest50_coco + +
faster_rcnn_fpn_syncbn_resnet101_v1d_coco + Pretrained model for faster_rcnn_fpn_syncbn_resnet101_v1d_coco is not available.
faster_rcnn_fpn_syncbn_resnet50_v1b_coco + Pretrained model for faster_rcnn_fpn_syncbn_resnet50_v1b_coco is not available.
faster_rcnn_resnet101_v1d_coco + +
faster_rcnn_resnet101_v1d_custom + TypeError: faster_rcnn_resnet101_v1d_custom() missing 1 required positional argument: 'classes'
faster_rcnn_resnet101_v1d_voc + Pretrained model for faster_rcnn_resnet101_v1d_voc is not available.
faster_rcnn_resnet50_v1b_coco + +
faster_rcnn_resnet50_v1b_custom + TypeError: faster_rcnn_resnet50_v1b_custom() missing 1 required positional argument: 'classes'
faster_rcnn_resnet50_v1b_voc + +
ssd_300_mobilenet0.25_coco + Pretrained model for ssd_300_mobilenet0.25_coco is not available.
ssd_300_mobilenet0.25_custom + TypeError: ssd_300_mobilenet0_25_custom() missing 1 required positional argument: 'classes'
ssd_300_mobilenet0.25_voc + Pretrained model for ssd_300_mobilenet0.25_voc is not available.
ssd_300_mobilenet1.0_lite_coco + Pretrained model for ssd_300_mobilenet1.0_coco is not available.
ssd_300_resnet34_v1b_coco + +
ssd_300_resnet34_v1b_custom + TypeError: ssd_300_resnet34_v1b_custom() missing 1 required positional argument: 'classes'
ssd_300_resnet34_v1b_voc + Pretrained model for ssd_300_resnet34_v1b_voc is not available.
ssd_300_vgg16_atrous_coco + +
ssd_300_vgg16_atrous_custom + TypeError: ssd_300_vgg16_atrous_custom() missing 1 required positional argument: 'classes'
ssd_300_vgg16_atrous_voc + +
ssd_300_vgg16_atrous_voc_int8 + -
ssd_512_mobilenet1.0_coco + +
ssd_512_mobilenet1.0_custom + TypeError: ssd_512_mobilenet1_0_custom() missing 1 required positional argument: 'classes'
ssd_512_mobilenet1.0_voc + +
ssd_512_mobilenet1.0_voc_int8 + -
ssd_512_resnet101_v2_voc + +
ssd_512_resnet152_v2_voc + Pretrained model for ssd_512_resnet152_v2_voc is not available.
ssd_512_resnet18_v1_coco + Pretrained model for ssd_512_resnet18_v1_coco is not available.
ssd_512_resnet18_v1_voc + Pretrained model for ssd_512_resnet18_v1_voc is not available.
ssd_512_resnet50_v1_coco + +
ssd_512_resnet50_v1_custom + TypeError: ssd_512_resnet50_v1_custom() missing 1 required positional argument: 'classes'
ssd_512_resnet50_v1_voc + +
ssd_512_resnet50_v1_voc_int8 + -
ssd_512_vgg16_atrous_coco + +
ssd_512_vgg16_atrous_custom + TypeError: ssd_512_vgg16_atrous_custom() missing 1 required positional argument: 'classes'
ssd_512_vgg16_atrous_voc + +
ssd_512_vgg16_atrous_voc_int8 + -
yolo3_darknet53_coco + +
yolo3_darknet53_custom + TypeError: yolo3_darknet53_custom() missing 1 required positional argument: 'classes'
yolo3_darknet53_voc + +
yolo3_mobilenet0.25_coco + Pretrained model for yolo3_mobilenet0.25_coco is not available.
yolo3_mobilenet0.25_custom + TypeError: yolo3_mobilenet0_25_custom() missing 1 required positional argument: 'classes'
yolo3_mobilenet0.25_voc + Pretrained model for yolo3_mobilenet0.25_voc is not available.
yolo3_mobilenet1.0_coco + +
yolo3_mobilenet1.0_custom + TypeError: yolo3_mobilenet1_0_custom() missing 1 required positional argument: 'classes'
yolo3_mobilenet1.0_voc + +

Semantic segmentation

Model Availability in OMZ (2023.02.24) Availability in the validation table
danet_resnet101_citys + +
danet_resnet50_citys + +
deeplab_resnest101_ade + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnest200_ade + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnest269_ade + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnest50_ade + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnet101_ade + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnet101_citys + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnet101_coco + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnet101_coco_int8 + -
deeplab_resnet101_voc + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnet101_voc_int8 + -
deeplab_resnet152_coco + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnet152_voc + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnet50_ade + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_resnet50_citys + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,256,60,60]
deeplab_v3b_plus_wideresnet_citys + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,60,60], got [1,256,80,80]
fastscnn_citys + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,16,16], got [1,32,32,32]
fcn_resnet101_ade + -
fcn_resnet101_coco + +
fcn_resnet101_coco_int8 + -
fcn_resnet101_voc + +
fcn_resnet101_voc_int8 + -
fcn_resnet50_ade + -
fcn_resnet50_voc + Pretrained model for fcn_resnet50_voc is not available.
hrnet_w18_small_v1_s + Pretrained model for hrnet_w18_small_v1_seg is not available.
hrnet_w18_small_v2_s + Pretrained model for hrnet_w18_small_v2_seg is not available.
hrnet_w48_s + Pretrained model for hrnet_w48_seg is not available.
icnet_resnet50_citys + MXNetError: Check failed: l == 1
icnet_resnet50_mhpv1 + MXNetError: Check failed: l == 1
nasnet_4_1056 + Pretrained model for nasnet_4_1056 is not available.
nasnet_5_1538 + Pretrained model for nasnet_5_1538 is not available.
nasnet_6_4032 + Pretrained model for nasnet_6_4032 is not available.
nasnet_7_1920 + Pretrained model for nasnet_7_1920 is not available.
psp_resnet101_ade + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,512,60,60]
psp_resnet101_citys + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,512,60,60]
psp_resnet101_coco + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,512,60,60]
psp_resnet101_coco_int8 + -
psp_resnet101_voc + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,512,60,60]
psp_resnet101_voc_int8 + -
psp_resnet50_ade + MXNetError: Check failed: shape_assign(&(*in_shape)[i], dshape): Incompatible input shape: expected [1,-1,64,64], got [1,512,60,60]

Instance segmentation

Model Availability in OMZ (2023.02.24) Availability in the validation table
custom_mask_rcnn_fpn + -
mask_rcnn_fpn_resnet101_v1d_coco + -
mask_rcnn_fpn_resnet18_v1b_coco + -
mask_rcnn_fpn_resnet50_v1b_coco + -
mask_rcnn_fpn_syncbn_mobilenet1_0_coco + -
mask_rcnn_fpn_syncbn_resnet18_v1b_coco + -
mask_rcnn_resnet101_v1d_coco + -
mask_rcnn_resnet18_v1b_coco + -
mask_rcnn_resnet50_v1b_coco + -

Pose estimation

Model Availability in OMZ (2023.02.24) Availability in the validation table
alpha_pose_resnet101_v1b_coco + -
mobile_pose_mobilenet1.0 + -
mobile_pose_mobilenetv2_1.0 + -
mobile_pose_mobilenetv3_large + -
mobile_pose_mobilenetv3_small + -
mobile_pose_resnet18_v1b + -
mobile_pose_resnet50_v1b + -
simple_pose_resnet101_v1b + -
simple_pose_resnet101_v1b_int8 + -
simple_pose_resnet101_v1d + -
simple_pose_resnet101_v1d_int8 + -
simple_pose_resnet152_v1b + -
simple_pose_resnet152_v1d + -
simple_pose_resnet18_v1b + -
simple_pose_resnet18_v1b_int8 + -
simple_pose_resnet50_v1b + -
simple_pose_resnet50_v1b_int8 + -
simple_pose_resnet50_v1d + -
simple_pose_resnet50_v1d_int8 + -

Action recognition

Model Availability in OMZ (2023.02.24) Availability in the validation table
c3d_kinetics400 + -
i3d_inceptionv1_kinetics400 + -
i3d_inceptionv3_kinetics400 + -
i3d_nl10_resnet101_v1_kinetics400 + -
i3d_nl10_resnet50_v1_kinetics400 + -
i3d_nl5_resnet101_v1_kinetics400 + -
i3d_nl5_resnet50_v1_kinetics400 + -
i3d_resnet101_v1_kinetics400 + -
i3d_resnet50_v1_custom + -
i3d_resnet50_v1_hmdb51 + -
i3d_resnet50_v1_kinetics400 + -
i3d_resnet50_v1_sthsthv2 + -
i3d_resnet50_v1_ucf101 + -
i3d_slow_resnet101_f16s4_kinetics700 + -
inceptionv1_hmdb51 + -
inceptionv1_kinetics400 + -
inceptionv1_sthsthv2 + -
inceptionv1_ucf101 + -
inceptionv3_hmdb51 + -
inceptionv3_kinetics400 + -
inceptionv3_kinetics400_int8 + -
inceptionv3_sthsthv2 + -
inceptionv3_ucf101 + -
inceptionv3_ucf101_int8 + -
p3d_resnet101_kinetics400 + -
p3d_resnet50_kinetics400 + -
r2plus1d_resnet101_kinetics400 + -
r2plus1d_resnet152_kinetics400 + -
r2plus1d_resnet18_kinetics400 + -
r2plus1d_resnet34_kinetics400 + -
r2plus1d_resnet50_kinetics400 + -
resnet101_v1b_kinetics400 + -
resnet101_v1b_sthsthv2 + -
resnet152_v1b_kinetics400 + -
resnet152_v1b_sthsthv2 + -
resnet18_v1b_kinetics400 + -
resnet18_v1b_kinetics400_int8 + -
resnet18_v1b_sthsthv2 + -
resnet34_v1b_kinetics400 + -
resnet34_v1b_sthsthv2 + -
resnet50_v1b_hmdb51 + -
resnet50_v1b_kinetics400 + -
resnet50_v1b_kinetics400_int8 + -
resnet50_v1b_sthsthv2 + -
resnet50_v1b_ucf101 + -
slowfast_16x8_resnet101_50_50_kinetics400 + -
slowfast_16x8_resnet101_kinetics400 + -
slowfast_4x16_resnet101_kinetics400 + -
slowfast_4x16_resnet50_custom + -
slowfast_4x16_resnet50_kinetics400 + -
slowfast_8x8_resnet101_kinetics400 + -
slowfast_8x8_resnet50_kinetics400 + -
vgg16_hmdb51 + -
vgg16_kinetics400 + -
vgg16_sthsthv2 + -
vgg16_ucf101 + -
vgg16_ucf101_int8 + -

Depth prediction

Model Availability in OMZ (2023.02.24) Availability in the validation table
monodepth2_resnet18_kitti_mono_640x192 + -
monodepth2_resnet18_kitti_mono_stereo_640x192 + -
monodepth2_resnet18_kitti_stereo_640x192 + -
monodepth2_resnet18_posenet_kitti_mono_640x192 + -
monodepth2_resnet18_posenet_kitti_mono_stereo_640x192 + -

Image classification on Cifar-10

Model Availability in OMZ (2023.02.24) Availability in the validation table
cifar_residualattentionnet452 + -
cifar_residualattentionnet56 + -
cifar_residualattentionnet92 + -
cifar_resnet110_v1 + -
cifar_resnet110_v2 + -
cifar_resnet20_v1 + -
cifar_resnet20_v2 + -
cifar_resnet56_v1 + -
cifar_resnet56_v2 + -
cifar_resnext29_16x64d + -
cifar_resnext29_32x4d + -
cifar_wideresnet16_10 + -
cifar_wideresnet28_10 + -
cifar_wideresnet40_8 + -

Object tracking

Model Availability in OMZ (2023.02.24) Availability in the validation table
siamrpn_alexnet_v2_otb15 + -