index.md

Overview of OpenVINO™ Toolkit Public Models

OpenVINO™ toolkit provides a set of public models that you can use for learning and demo purposes or for developing deep learning software. Most recent version is available in the repo on Github.

The models can be downloaded via Model Downloader (<OPENVINO_INSTALL_DIR>/deployment_tools/open_model_zoo/tools/downloader).

Classification

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
AlexNet	Caffe*	alexnet	56.598%/79.812%	1.5	60.965
CaffeNet	Caffe*	caffenet	56.714%/79.916%	1.5	60.965
DenseNet 121	Caffe* TensorFlow* Caffe2*	densenet-121 densenet-121-tf densenet-121-caffe2	74.42%/92.136% 74.29%/91.98% 74.904%/92.192%	5.289~5.724	7.971
DenseNet 161	Caffe* TensorFlow*	densenet-161 densenet-161-tf	77.55%/93.92% 76.446%/93.228%	14.128~15.561	28.666
DenseNet 169	Caffe* TensorFlow*	densenet-169 densenet-169-tf	76.106%/93.106% 75.76%/92.81%	6.16~6.788	14.139
DenseNet 201	Caffe*	densenet-201	76.886%/93.556%	8.673	20.001
EfficientNet B0	TensorFlow* PyTorch*	efficientnet-b0 efficientnet-b0-pytorch	75.70%/92.76% 76.91%/93.21%	0.819	5.268
EfficientNet B0 AutoAugment	TensorFlow*	efficientnet-b0_auto_aug	76.43%/93.04%	0.819	5.268
EfficientNet B5	TensorFlow* PyTorch*	efficientnet-b5 efficientnet-b5-pytorch	83.33%/96.67% 83.69%/96.71%	21.252	30.303
EfficientNet B7	PyTorch*	efficientnet-b7-pytorch	84.42%/96.91%	77.618	66.193
EfficientNet B7 AutoAugment	TensorFlow*	efficientnet-b7_auto_aug	84.68%/97.09%	77.618	66.193
HBONet 1.0	PyTorch*	hbonet-1.0	73.1%/91.0%	0.6208	4.5443
HBONet 0.5	PyTorch*	hbonet-0.5	67.0%/86.9%	0.1977	2.5287
HBONet 0.25	PyTorch*	hbonet-0.25	57.3%/79.8%	0.0758	1.9299
Inception (GoogleNet) V1	Caffe* TensorFlow*	googlenet-v1 googlenet-v1-tf	68.928%/89.144% 69.814%/89.6%	3.016~3.266	6.619~6.999
Inception (GoogleNet) V2	Caffe* TensorFlow*	googlenet-v2 googlenet-v2-tf	72.024%/90.844% 74.084%/91.798%	4.058	11.185
Inception (GoogleNet) V3	TensorFlow* PyTorch*	googlenet-v3 googlenet-v3-pytorch	77.904%/93.808% 77.696%/93.696%	11.469	23.817
Inception (GoogleNet) V4	TensorFlow*	googlenet-v4-tf	80.204%/95.21%	24.584	42.648
Inception-ResNet V2	TensorFlow*	inception-resnet-v2-tf	80.14%/95.10%	22.227	30.223
MobileNet V1 0.25 128	Caffe*	mobilenet-v1-0.25-128	40.54%/65%	0.028	0.468
MobileNet V1 0.5 160	Caffe*	mobilenet-v1-0.50-160	59.86%/82.04%	0.156	1.327
MobileNet V1 0.5 224	Caffe*	mobilenet-v1-0.50-224	63.042%/84.934%	0.304	1.327
MobileNet V1 1.0 224	Caffe* TensorFlow*	mobilenet-v1-1.0-224 mobilenet-v1-1.0-224-tf	69.496%/89.224% 71.03%/89.94%	1.148	4.221
MobileNet V2 1.0 224	Caffe* TensorFlow* PyTorch*	mobilenet-v2 mobilenet-v2-1.0-224 mobilenet-v2-pytorch	71.218%/90.178% 70.20%/89.61% 71.81%/90.396%	0.615~0.876	3.489
MobileNet V2 1.4 224	TensorFlow*	mobilenet-v2-1.4-224	74.09%/91.97%	1.183	6.087
MobileNet V3 Small 1.0	TensorFlow*	mobilenet-v3-small-1.0-224-tf	67.36%/87.45%	0.121	2.537
MobileNet V3 Large 1.0	TensorFlow*	mobilenet-v3-large-1.0-224-tf	75.70%/92.76%	0.4536	5.4721
DenseNet 121, alpha=0.125	MXNet*	octave-densenet-121-0.125	76.066%/93.044%	4.883	7.977
ResNet 26, alpha=0.25	MXNet*	octave-resnet-26-0.25	76.076%/92.584%	3.768	15.99
ResNet 50, alpha=0.125	MXNet*	octave-resnet-50-0.125	78.19%/93.862%	7.221	25.551
ResNet 101, alpha=0.125	MXNet*	octave-resnet-101-0.125	79.182%/94.42%	13.387	44.543
ResNet 200, alpha=0.125	MXNet*	octave-resnet-200-0.125	79.99%/94.866%	25.407	64.667
ResNeXt 50, alpha=0.25	MXNet*	octave-resnext-50-0.25	78.772%/94.18%	6.444	25.02
ResNeXt 101, alpha=0.25	MXNet*	octave-resnext-101-0.25	79.556%/94.444%	11.521	44.169
SE-ResNet 50, alpha=0.125	MXNet*	octave-se-resnet-50-0.125	78.706%/94.09%	7.246	28.082
open-closed-eye-0001	PyTorch*	open-closed-eye-0001	95.84%	0.0014	0.0113
ResNet 18	PyTorch*	resnet-18-pytorch	69.754%/89.088%	3.637	11.68
ResNet 34	PyTorch*	resnet-34-pytorch	73.30%/91.42%	7.3409	21.7892
ResNet 50	Caffe* PyTorch* Caffe2* TensorFlow*	resnet-50 resnet-50-pytorch resnet-50-caffe2 resnet-50-tf	75.168%/92.212% 76.128%/92.858% 76.38%/93.188% 76.17%/92.98%	6.996~8.216	25.53
ResNet 101	Caffe*	resnet-101	76.364%/92.902%	14.441	44.496
ResNet 152	Caffe*	resnet-152	76.802%/93.192%	21.89	60.117
SE-Inception	Caffe*	se-inception	75.996%/92.964%	4.091	11.922
SE-ResNet 50	Caffe*	se-resnet-50	77.596%/93.85%	7.775	28.061
SE-ResNet 101	Caffe*	se-resnet-101	78.252%/94.206%	15.239	49.274
SE-ResNet 152	Caffe*	se-resnet-152	78.506%/94.45%	22.709	66.746
SE-ResNeXt 50	Caffe*	se-resnext-50	78.968%/94.63%	8.533	27.526
SE-ResNeXt 101	Caffe*	se-resnext-101	80.168%/95.19%	16.054	48.886
SqueezeNet v1.0	Caffe*	squeezenet1.0	57.684%/80.38%	1.737	1.248
SqueezeNet v1.1	Caffe* Caffe2*	squeezenet1.1 squeezenet1.1-caffe2	58.382%/81% 56.502%/79.576%	0.785	1.236
VGG 16	Caffe*	vgg16	70.968%/89.878%	30.974	138.358
VGG 19	Caffe* Caffe2*	vgg19 vgg19-caffe2	71.062%/89.832% 71.062%/89.832%	39.3	143.667

Segmentation

Semantic segmentation is an extension of object detection problem. Instead of returning bounding boxes, semantic segmentation models return a "painted" version of the input image, where the "color" of each pixel represents a certain class. These networks are much bigger than respective object detection networks, but they provide a better (pixel-level) localization of objects and they can detect areas with complex shape.

Semantic Segmentation

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
DeepLab V3	TensorFlow*	deeplabv3	66.85%	11.469	23.819

Instance Segmentation

Instance segmentation is an extension of object detection and semantic segmentation problems. Instead of predicting a bounding box around each object instance instance segmentation model outputs pixel-wise masks for all instances.

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
Mask R-CNN Inception ResNet V2	TensorFlow*	mask_rcnn_inception_resnet_v2_atrous_coco	39.8619%/35.3628%	675.314	92.368
Mask R-CNN Inception V2	TensorFlow*	mask_rcnn_inception_v2_coco	27.1199%/21.4805%	54.926	21.772
Mask R-CNN ResNet 50	TensorFlow*	mask_rcnn_resnet50_atrous_coco	29.7512%/27.4597%	294.738	50.222
Mask R-CNN ResNet 101	TensorFlow*	mask_rcnn_resnet101_atrous_coco	34.9191%/31.301%	674.58	69.188

3D Semantic Segmentation

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
Brain Tumor Segmentation	MXNet*	brain-tumor-segmentation-0001	92.4003%	409.996	38.192
Brain Tumor Segmentation 2	PyTorch*	brain-tumor-segmentation-0002	91.4826%	300.801	4.51

Object Detection

Several detection models can be used to detect a set of the most popular objects - for example, faces, people, vehicles. Most of the networks are SSD-based and provide reasonable accuracy/performance trade-offs.

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
CTPN	TensorFlow*	ctpn	1.0864%	55.813	17.237
CenterNet (CTDET with DLAV0) 384x384	ONNX*	ctdet_coco_dlav0_384	41.6105%	34.994	17.911
CenterNet (CTDET with DLAV0) 512x512	ONNX*	ctdet_coco_dlav0_512	44.2756%	62.211	17.911
FaceBoxes	PyTorch*	faceboxes-pytorch	83.565%	1.8975	1.0059
Faster R-CNN with Inception-ResNet v2	TensorFlow*	faster_rcnn_inception_resnet_v2_atrous_coco	36.76%/52.41%	30.687	13.307
Faster R-CNN with Inception v2	TensorFlow*	faster_rcnn_inception_v2_coco	25.65%/40.04%	30.687	13.307
Faster R-CNN with ResNet 50	TensorFlow*	faster_rcnn_resnet50_coco	27.47%/42.87%	57.203	29.162
Faster R-CNN with ResNet 101	TensorFlow*	faster_rcnn_resnet101_coco	30.95%/47.21%	112.052	48.128
MobileFace Detection V1	MXNet*	mobilefacedet-v1-mxnet	78.7488%	3.5456	7.6828
MTCNN	Caffe*	mtcnn: mtcnn-p mtcnn-r mtcnn-o	48.1308%/62.2625%	3.3715 0.0031 0.0263	0.0066 0.1002 0.3890
Pelee	Caffe*	pelee-coco	21.9761%	1.290	5.98
RetinaNet with Resnet 50	TensorFlow*	retinanet-tf	33.15%	238.9469	64.9706
R-FCN with Resnet-101	TensorFlow*	rfcn-resnet101-coco-tf	28.40%/45.02%	53.462	171.85
SSD 300	Caffe*	ssd300	85.0791%	62.815	26.285
SSD 512	Caffe*	ssd512	90.3845%	180.611	27.189
SSD with MobileNet	Caffe* TensorFlow*	mobilenet-ssd ssd_mobilenet_v1_coco	79.8377% 23.32%	2.316~2.494	5.783~6.807
SSD with MobileNet FPN	TensorFlow*	ssd_mobilenet_v1_fpn_coco	35.5453%	123.309	36.188
SSD with MobileNet V2	TensorFlow*	ssd_mobilenet_v2_coco	24.9452%	3.775	16.818
SSD lite with MobileNet V2	TensorFlow*	ssdlite_mobilenet_v2	24.2946%	1.525	4.475
SSD with ResNet-50 V1 FPN	TensorFlow*	ssd_resnet50_v1_fpn_coco	38.4557%	178.6807	59.9326
SSD with ResNet 34 1200x1200	PyTorch*	ssd-resnet34-1200-onnx	20.7198%/39.2752%	433.411	20.058
RetinaFace-R50	MXNet*	retinaface-resnet50	87.2902%	100.8478	29.427
RetinaFace-Anti-Cov	MXNet*	retinaface-anti-cov	77.1531%	2.7781	0.5955
YOLO v1 Tiny	TensorFlow.js*	yolo-v1-tiny-tf	72.1716%	6.9883	15.8587
YOLO v2 Tiny	TensorFlow.js*	yolo-v2-tiny-tf	27.3443%/29.1184%	5.4236	11.2295
YOLO v2	Keras*	yolo-v2-tf	53.1453%/56.483%	63.0301	50.9526
YOLO v3	Keras*	yolo-v3-tf	62.2759%/67.7221%	65.9843	61.9221

Face Recognition

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
FaceNet	TensorFlow*	facenet-20180408-102900	98.4522%	2.846	23.469
LResNet34E-IR,ArcFace@ms1m-refine-v1	MXNet*	face-recognition-resnet34-arcface	98.7488%	8.934	34.129
LResNet50E-IR,ArcFace@ms1m-refine-v1	MXNet*	face-recognition-resnet50-arcface	98.8835%	12.637	43.576
LResNet100E-IR,ArcFace@ms1m-refine-v2	MXNet*	face-recognition-resnet100-arcface	99.0218%	24.209	65.131
MobileFaceNet,ArcFace@ms1m-refine-v1	MXNet*	face-recognition-mobilefacenet-arcface	98.8695%	0.449	0.993
SphereFace	Caffe*	Sphereface	98.8321%	3.504	22.671

Human Pose Estimation

Human pose estimation task is to predict a pose: body skeleton, which consists of keypoints and connections between them, for every person in an input image or video. Keypoints are body joints, i.e. ears, eyes, nose, shoulders, knees, etc. There are two major groups of such metods: top-down and bottom-up. The first detects persons in a given frame, crops or rescales detections, then runs pose estimation network for every detection. These methods are very accurate. The second finds all keypoints in a given frame, then groups them by person instances, thus faster than previous, because network runs once.

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
human-pose-estimation-3d-0001	PyTorch*	human-pose-estimation-3d-0001	100.44437mm	18.998	5.074
single-human-pose-estimation-0001	PyTorch*	single-human-pose-estimation-0001	69.0491%	60.125	33.165

Monocular Depth Estimation

The task of monocular depth estimation is to predict a depth (or inverse depth) map based on a single input image. Since this task contains - in the general setting - some ambiguity, the resulting depth maps are often only defined up to an unknown scaling factor.

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
midasnet	PyTorch*	midasnet	7.5878	207.4915	104.0814

Image Inpainting

Image inpainting task is to estimate suitable pixel information to fill holes in images.

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
GMCNN Inpainting	TensorFlow*	gmcnn-places2-tf	33.47Db	691.1589	12.7773

Style Transfer

Style transfer task is to transfer the style of one image to another.

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
fast-neural-style-mosaic-onnx	ONNX*	fast-neural-style-mosaic-onnx	12.04dB	15.518	1.679

Action Recognition

The task of action recognition is to predict action that is being performed on a short video clip (tensor formed by stacking sampled frames from input video).

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
RGB-I3D, pretrained on ImageNet*	TensorFlow*	i3d-rgb-tf	65.96%/86.01%	278.9815	12.6900

Colorization

Colorization task is to predict colors of scene from grayscale image.

Model Name	Implementation	OMZ Model Name	Accuracy	GFlops	mParams
colorization-v2	Caffe*	colorization-v2		64.0047	32.2354
colorization-v2-norebal	Caffe*	colorization-v2-norebal		64.0047	32.2354

Legal Information

[*] Other names and brands may be claimed as the property of others.