Represented deep models are available in Open Model Zoo.
Model | Availability in OMZ (2023.03.04) | Availability in the validation table |
---|---|---|
densenet-121-tf | +- | - (Error parsing message with type 'tensorflow.GraphDef') |
efficientnet-b0 | +- | - ('NoneType' object is not iterable) |
googlenet-v1-tf (inceptionv1) | + | - |
googlenet-v2-tf (inceptionv2) | + | - |
googlenet-v3 (inceptionv3) | + | - |
googlenet-v4-tf (inceptionv4) | + | - |
inception-resnet-v2-tf | + | + |
mixnet-l | + | - ('NoneType' object is not iterable) |
mobilenet-v1-1.0-224-tf | + | + |
mobilenet-v2-1.0-224 | + | + |
mobilenet-v2-1.4-224 | + | + |
mobilenet-v3-small-1.0-224-tf | +- | - (Error parsing message with type 'tensorflow.GraphDef') |
mobilenet-v3-large-1.0-224-tf | +- | - (Error parsing message with type 'tensorflow.GraphDef') |
resnet-50-tf | + | + |
Notes:
-
Inference implementation for GoogleNet-models supported for batch size that equals 1.
-
Inference of densenet-121-tf, efficientnet-b0, mobilenet-v3-*, mixnet-l fails.
-
"+-" in the column of availability in OMZ (2023.03.04) means that model was converted by
omz_converter
from.ckpt
or.h5
into.pb
.omz_downloader --name <model_name> # export is required for GoogleNet-models export PYTHONPATH=`pwd`:`pwd`/public/<model_name>/models/research/slim omz_converter --name <model_name>
Model | Availability in OMZ (2023.03.04) | Availability in the validation table |
---|---|---|
ctpn | + | - |
efficientdet-d0 | + | - |
efficientdet-d1 | + | - |
faster_rcnn_inception_resnet_v2_atrous_coco | + | - |
faster_rcnn_resnet50_coco | + | - |
retinanet | + | - |
rfcn-resnet101-coco | + | - |
ssd_mobilenet_v1_coco | + | - |
ssd_mobilenet_v1_fpn_coco | + | - |
ssdlite_mobilenet_v2 | + | - |
Model | Availability in OMZ (2023.03.04) | Availability in the validation table |
---|---|---|
deeplabv3 | + | - |
Model | Availability in OMZ (2023.03.04) | Availability in the validation table |
---|---|---|
mask_rcnn_resnet50_atrous_coco | + | - |
mask_rcnn_inception_resnet_v2_atrous_coco | + | - |
There are deep models for solving non-classical computer vision tasks. The complete list is available here.